Genetic Diversity of Newcastle Disease Virus in Wild Birds and Pigeons in West Africa

Abstract

In West and Central Africa, virulent Newcastle disease virus (NDV) strains of the recently identified genotypes XIV, XVII, and XVIII are enzootic in poultry, representing a considerable threat to the sector. The increasing number of reports of virulent strains in wild birds at least in other parts of the world raised the question of a potential role of wild birds in the spread of virulent NDV in sub-Saharan Africa as well. We investigated 1,723 asymptomatic birds sampled at live-bird markets and sites important for wild-bird conservation in Nigeria and 19 sick or dead wild birds in Côte d'Ivoire for NDV class I and II. Typical avirulent wild-type genotype I strains were found in wild waterfowl in wetlands in northeastern Nigeria. They were unrelated to vaccine strains, and the involvement of inter- or intracontinental migratory birds in their circulation in the region is suggested. Phylogenetic analyses also revealed that genotype VI strains found in pigeons, including some putative new subgenotype VIh and VIi strains, were introduced on multiple separate occasions in Nigeria. A single virulent genotype XVIII strain was found in a dead wild bird in Côte d'Ivoire, probably as a result of spillover from sick poultry. In conclusion, screening of wild birds and pigeons for NDV revealed the presence a variety of virulent and avirulent strains in West Africa but did not provide strong evidence that wild birds play an important role in the spread of virulent strains in the region.

INTRODUCTION

Newcastle disease, caused by virulent Newcastle disease virus (NDV), an avian Paramyxovirus, is one of the most important diseases in poultry. Depending on the severity of the clinical symptoms observed in chickens, NDV are separated into asymptomatic enteric, lentogenic, mesogenic, viscerotropic velogenic, and neurotropic velogenic pathotypes (1). In addition, avirulent and virulent strains can be distinguished on the basis of the cleavage site sequence of their fusion (F) protein (1), one of the most important virulence factors.

Although all NDV strains belong to a single serotype, they are nevertheless genetically highly diverse. Based on the complete fusion gene sequences, NDV strains are divided into two classes, I and II. In the latter, at least 18 genotypes and multiple subgenotypes have been defined (2–4), but the diversity continues to increase as surveillance improves.

Virtually all domestic and wild bird species are susceptible to infection with NDV (5). Wild waterbirds seem to be the reservoir of avirulent strains, whereas poultry are the most likely reservoir of virulent viruses, but both hosts exchange viruses. In Israel and Mexico, when captive zoo birds became infected with virulent strains similar to those causing outbreaks in local poultry, spillover from infected domestic birds was suspected but never demonstrated (6, 7). However, free-living migratory species, such as waterfowls or white storks, may carry virulent NDV strains without obvious contact with poultry (8–10). Thus, it could be that wild birds are carriers of virulent strains but transmission routes of virulent NDV strains in particular are not yet fully understood. In West and Central Africa, virulent strains of the recently identified genotypes XIV, XVII, and XVIII are enzootic in poultry (4). Highly similar strains in poultry across West and Central Africa (4) and an increasing number of reports of virulent strains in wild birds on other continents (8–10) raise the question of the potential role of wild birds in the spread of virulent NDV in sub-Saharan Africa as well. Anti-NDV antibodies and NDV strains have been detected in wild bird species in South Africa (11), Burkina Faso (12), and Nigeria (13, 14), but genetic information was available only from doves and pigeons (genotype VI) in Nigeria (15, 16) and South Africa (17). In this study, we therefore investigated the presence of NDV in wild birds and pigeons in Nigeria and Côte d'Ivoire.

MATERIALS AND METHODS

Sample collection.

A total of 1,723 asymptomatic birds were sampled in six Nigerian states (Oyo, n = 830; Yobe, n = 668; Plateau, n = 154; Sokoto, n = 37; Lagos, n = 32; Nasarawa, n = 2) between 2006 and 2013 (Table 1; Fig. 1). Pooled oropharyngeal and cloacal swabs (n = 1345), cloacal swabs (n = 74), tracheal swabs (n = 12), and fresh feces (n = 292) were collected, and bird species were recorded. Birds were sampled at live-bird markets (n = 140) or in three sites important for wild bird conservation (n = 1,583), including the International Institute for Tropical Agriculture Forest Reserve (Oyo State), the Amurum woodlands (Plateau State), and the Hadejia-Nguru wetlands (Yobe State). Organs from 19 moribund or dead birds were collected in Lagunes province in Côte d'Ivoire between 2006 and 2008 (Table 1).

Table 1.

Number of birds (classified by order and family) sampled in six Nigerian states (2006 to 2013) and Côte d'Ivoire (2006 to 2008) tested for NDV

Order	Family	No. of samples collected (no. positive)
		Nigeria						Côte d'Ivoire
		Oyo	Yobe	Plateau	Lagos	Nasarawa	Sokoto
Accipitriformes	Accipitridae	4	4
Anseriformes	Anatidae	143	130 (5)
Charadriiformes	Burhinidae	3
	Charadriidae	6	1
	Jacanidae	8	32
	Rostratulidae	12
	Scolopacidae		18
Ciconiiformes	Ciconiidae		6
Coliiformes	Coliidae			9
Columbiformes	Columbidae	70 (8)	161	4	32 (1)	2	37	11
Coraciiformes	Alcedinidae	10	20	1
	Coraciidae		1
Cuculiformes	Cuculidae		1	1
Galliformes	Phasianidae		1					1
Gruiformes	Rallidae	5	8
Passeriformes	Acrocephalidae		4	1
	Alaudidae			1
	Cisticolidae	46		22
	Corvidae							4
	Dicruridae	2	1
	Emberizidae			1
	Estrildidae	23	14	38
	Fringillidae		37
	Hirundinidae		9
	Macrosphenidae	8		6
	Malconotidae			7
	Monarchidae	28
	Motacillidae		14	1
	Muscicapidae	14		6
	Nectariniidae	82		8
	Nicatoridae	5
	Passeridae		56	2
	Pellorneidae	9
	Phylloscoidae			2
	Platysteiridae	11		1
	Ploceidae	34	123	12				1 (1)
	Pycnonotidae	225		3
	Stenostiridae			1
	Sylviidae			2
	Turdidae	6		2
	Viduidae		21	1
	Zosteropidae			4
	Unknown	3
Pelecaniformes	Ardeidae		5					1
Piciformes	Indicatoridae		1	2
	Lybiidae	2		11
	Picidae	5		5
Unknown		66						1
Total		830 (8)	668 (5)	154	32 (1)	2	37	19 (1)

Fig 1.

Collection sites in Nigeria and Côte d'Ivoire. Sampled provinces or states are indicated by circles. The number of positive samples and the total number of samples tested are in parentheses. (Base map copyright, Map Resources; reproduced with permission.)

Nucleic acid extraction, PCR, and sequencing.

All swabs and approximately 100 mg of fecal material were homogenized and stored in 500 μl of virus transport medium (phosphate-buffered saline with 2,000 U of penicillin/ml, 200 mg of streptomycin/ml, 2,000 U of polymyxin B/ml, 250 mg of gentamicin/ml, 60 mg of ofloxacin/ml, 200 mg of sulfamethoxazole/ml, and 2.5 mg of amphotericin B/ml). RNA was extracted from 140 μl of medium using a QIAamp viral RNA minikit (Qiagen, Venlo, The Netherlands) or from 50 μl using a MagMAX-96 AI/ND viral RNA isolation kit (Life Technologies, Merelbeke, Belgium) with KingFisher (Thermo Fisher, Waltham, MA). Organs were homogenized in 600 μl of lysis buffer of the RNeasy minikit (Qiagen) with stainless steel beads and TissueLyzer II (Qiagen). RNA extraction was then performed according to the manufacturer's instructions. NDV detection was performed by a multiplex real-time RT-PCR detecting both class I and class II strains (18). Full-length (n = 13) or partial (n = 2) F genes were amplified using several overlapping fragments in (semi)nested PCRs (primer sequences are available upon request). PCR product purification, sequencing, and contig assembly were performed as described before (4). The primers used for generating the PCR fragments were also utilized in the sequencing reaction.

Data analysis.

Genetic distances and between-group distances were calculated using the maximum composite likelihood model and 500 bootstrap replicates as implemented in MEGA v5.03 (19). Amino acid distances of deduced fusion protein sequences were calculated with the Poisson correction model (MEGA v5.03). Phylogenetic analyses including 1,164 complete F gene sequences available on GenBank and 13 sequences obtained in this study were performed with the maximum likelihood method (MEGA v5.03). Criteria to define new subgenotypes were based on those reported previously (3, 4), i.e., (i) the tree topology, (ii) bootstrap values of ≥60%, (iii) mean evolutionary distances between subgenotypes between 0.03 and 0.1, and (iv) a minimum of four sequences from at least two distinct outbreaks per subgenotype. Phylogenetic analyses on partial sequences (372 nucleotides), including 3,453 publicly available sequences and 15 sequences from this study, were performed with the neighbor-joining method with the Kimura 2-parameter model (MEGA v5.03). Representative strains were selected on the basis of these preliminary analyses. Final trees were calculated with the maximum likelihood method and the general time-reversible (GTR) substitution model with a gamma (G) and invariant (I) site heterogeneity model and are displayed in Fig. 2. The following strain nomenclature was used: host/country/strain number/year.

Fig 2.

Phylogenetic analysis of complete (1,662 nt) (A) or partial (372 nt) (B) F gene sequences. (A) Phylogenetic tree calculated with class I strain Goose/Alaska/415/1991 as the outgroup. Sequences generated in this study are in bold. Previously published sequences are indicated with their accession numbers. African strains are shown with black circles. (B) The state of origin of Nigerian sequences is indicated as follows: gray circles, Jigawa State; white circles, Lagos State; black squares, Oyo State; gray squares, Kano State; black triangles, Yobe State; gray triangles, Sokoto State. Only bootstrap values of ≥70% are shown. The scale corresponds to the number of base substitutions per site. GWH, green wood hoopoe.

Nucleotide sequence accession numbers.

Sequences were submitted to GenBank under accession numbers HG326600 to HG326609 and HG424625 to HG424629.

RESULTS AND DISCUSSION

Fifteen out of 1,742 birds tested positive for NDV. Based on partial or full F-gene sequences (Fig. 2A and B), five strains were assigned to genotype I, nine strains to genotype VI, and one to genotype XVIII, all in class II. No sample was positive for class I strains.

Genotype I.

Five highly similar sequences (genetic distance, 0 to 0.1%) from spur-winged geese (Plectropterus gambensis, order Anseriformes) sampled in the Hadejia-Nguru wetlands in Yobe State, Nigeria, during three consecutive days in April 2008 formed a cluster within genotype I (Fig. 2A), suggesting that they shared a very recent ancestor. Their cleavage site sequence, ¹¹²GKQGR↓L¹¹⁷ (the arrow represents the site of cleavage of the precursor protein, F0, into its F1 and F2 subunits), was typical of avirulent genotype I strains. All strains belonged to a cluster of strains found mainly in migratory wild waterfowl and domestic ducks in Asia, the United States, and Europe (Fig. 2A). They were not related to live vaccine strains derived from Ulster, Queensland/V4, or I-2, in contrast to other genotype I strains found in South Africa, Nigeria, Cameroon, and Mozambique (Fig. 2B). Their antigenic distances from genotype I and II vaccine strains ranged from 0.011 to 0.035 and 0.069 to 0.075, respectively (Table 2). No similar genotype I strains were detected in poultry in West and Central Africa even during more intensive sampling periods in 2006 to 2011 (4), suggesting that these strains may have been introduced more recently by intercontinental migratory birds. Indeed, in the wetlands in northeastern Nigeria, spur-winged geese often mix with overwintering Eurasian migratory birds (20). Equally, the paucity of wild bird screening for NDV in Africa makes it possible that such genotype I strains circulate in African wild birds and may have been introduced in the wetlands by birds migrating within the African continent. However, virus introduction by poultry also cannot be excluded, despite the ban on importation of poultry into Nigeria since 2002. The presence of free-roaming domestic ducks in the wetlands (21) offers opportunities for transmission of NDV from wild to domestic waterfowl and vice versa.

Table 2.

Estimates of evolutionary distances of deduced fusion proteins between 13 strains obtained in this study and common vaccine strains of genotypes I and II^a

Strain	Distance (no. of amino acid substitutions per site)
	Genotype I			Genotype II
	Queensland-V4 (M24693)	Ulster (M24694)	I-2 (AY935499)	LaSota (AY845400)	B1 (JN872150)
Pigeon/Nigeria/NIE07-061/2007	0.101	0.087	0.099	0.115	0.109
Pigeon/Nigeria/NIE07-062/2007	0.101	0.087	0.099	0.115	0.109
Pigeon/Nigeria/NIE07-063/2007	0.103	0.089	0.101	0.117	0.111
Pigeon/Nigeria/NIE09-1898/2009	0.101	0.087	0.097	0.115	0.109
Pigeon/Nigeria/NIE13-005/2013	0.129	0.115	0.125	0.156	0.15
Pigeon/Nigeria/NIE13-008/2013	0.129	0.115	0.125	0.156	0.15
Pigeon/Nigeria/NIE13-092/2013	0.111	0.099	0.111	0.125	0.119
Pigeon/Nigeria/NIE13-093/2013	0.111	0.099	0.111	0.125	0.119
Spur-winged goose/Nigeria/NIE08-0121/2008	0.029	0.013	0.035	0.075	0.069
Spur-winged goose/Nigeria/NIE08-0124/2008	0.027	0.011	0.033	0.073	0.071
Spur-winged goose/Nigeria/NIE08-0232/2008	0.028	0.011	0.033	0.073	0.069
Spur-winged goose/Nigeria/NIE08-0273/2008	0.028	0.011	0.033	0.071	0.069
Village weaver/Ivory Coast/CIV08-032/2006	0.101	0.091	0.093	0.119	0.117

^aAnalyses were conducted using the Poisson correction model in MEGA v5.03. All ambiguous positions were removed for each sequence pair. There were a total of 553 positions in the final data set.

Interestingly, 650 birds belonging to 44 other species were also sampled during the same period of time (March-April 2008) in the same region, but no other species tested positive for NDV. Also, spur-winged geese and white-faced whistling ducks (Dendrocygna viduata) were the only two species carrying avian influenza H5N2 viruses among 17 (22) and 44 (23) species investigated, suggesting that they may play a particular role in the epidemiology of both avian influenza and NDV in the region. These two species have similar behavior, i.e., they forage in shallow water, are both intracontinental migrants, and gather around water bodies (24) where they can intermix with migratory birds, which have been shown as risk factors for avian influenza reservoirs (25).

Genotype VI.

Nine sequences from pigeons sampled in Nigerian live-bird markets clustered in genotype VI (Fig. 2A and B). One cluster was formed by three very similar strains (genetic distances 0.1 to 0.2%) collected in Oyo State in 2007 and a fourth sequence from a pigeon sampled in Lagos State in 2009 (genetic distance of 0.7 to 0.8% from the three strains from 2007) (Fig. 2A). They shared the same predicted virulent cleavage site sequence, ¹¹²RRKKR↓F¹¹⁷. Phylogenetic analyses based on partial F gene revealed that these four strains also clustered with two other Nigerian strains from Oyo (Parrot/Nigeria/NIE139/2007) and Sokoto (Pigeon/Nigeria/NIE95/2007) states (Fig. 2B). Three strains (Oyo State, 2013) clustered together with pigeon strains from Jigawa State (Fig. 2A and B) and shared the same virulent cleavage site, ¹¹²RRRKR↓F¹¹⁷. The clustering of the Nigerian strains together with Pigeon/Argentina/Capital_3/97 as a distinct subgroup within genotype VI and their great genetic distance from other VI subgenotypes (0.07 to 0.104) (Table 3), greater than 0.03 (the cutoff for subgenotypes), suggested the definition of a new subgenotype, VIh. In addition, two more strains (Oyo State, 2013) clustered with recent Italian strains identified during an episode of high mortality in collared doves (26) and shared the same ¹¹²RRQKR↓F¹¹⁷ cleavage site motif. Similarly, their genetic distances from other subgenotypes within VI (0.094 to 0.123) (Table 3) would indicate an additional subgenotype, VIi. However, some of the genetic distances between subgenotype VIi and the other VI subgenotypes were even higher than the 0.1 cutoff value between genotypes, suggesting that subgenotype VIi may need to be further subdivided when more strains become available. The presence of similar strains (subgenotype VIh) (Fig. 2B) in several Nigerian states suggests an exchange between regions. However, none of them were related to other genotype VI strains from the more distant countries, such as South Africa (subgenotype VIa), Kenya, or Ethiopia (Fig. 2B), suggesting multiple introductions of genotype VI strains into Africa during the last 10 years and limited long-distance circulation.

Table 3.

Estimates of evolutionary distances over sequence pairs between subgenotypes of genotype VI

Subgenotype	No. of strains	No. of base substitutions per site or standard error estimatea
		VIa	VIb	VIc	VIe	VIf	VIg	VIh	VIi
VIa	70		[0.005]	[0.006]	[0.006]	[0.006]	[0.007]	[0.005]	[0.008]
VIb	10	0.067		[0.005]	[0.006]	[0.005]	[0.006]	[0.006]	[0.008]
VIc	12	0.09	0.071		[0.006]	[0.006]	[0.006]	[0.007]	[0.007]
VIe	12	0.077	0.064	0.085		[0.005]	[0.008]	[0.006]	[0.008]
VIf	12	0.07	0.057	0.083	0.06		[0.007]	[0.005]	[0.008]
VIg	4	0.092	0.08	0.088	0.097	0.089		[0.008]	[0.008]
VIh	9	0.07	0.072	0.101	0.086	0.082	0.104		[0.009]
VIi	5	0.112	0.101	0.106	0.11	0.107	0.094	0.123

^aNumbers below the diagonal are numbers of base substitutions per site from averaging over all sequence pairs between groups (1,662 nucleotides). Numbers above the diagonal are standard error estimates (500 bootstrap replicates). Analyses were conducted using the maximum composite likelihood model in MEGA v5.03. Codon positions included were first + second + third + noncoding. All ambiguous positions were removed for each sequence pair.

All neutralizing epitopes on the F protein (D72, E74, A75, K78, A79, L343, ¹⁵¹ILRLKESIAATNEAVHEVTDG¹⁷¹, [27, 28]) were conserved between vaccine and genotype VI strains, except for two subgenotype VIh strains (Pigeon/Nigeria/NIE13-092/2013 and Pigeon/Nigeria/NIE13-093/2013) that shared a A79V substitution. Amino acid distances of genotype VI strains to genotype I or II vaccine strains ranged from 0.087 to 0.129 and 0.109 to 0.156, respectively (Table 2).

In some countries outside Africa, genotype VI strains from domestic pigeons spread to other species, such as feral pigeons and doves, where they became enzootic (29, 30), sometimes resulting in high mortality rates (26, 31). In our study, most NDV-positive pigeons were sampled in live-bird markets, indicating that they were most likely domestic pigeons reared for their meat. Nevertheless, 169 samples from nine wild dove species were negative for NDV, suggesting that genotype VI may not yet be enzootic in wild doves in the sampled region. Genotype VI was also not found in more than 3,000 domestic birds sampled in West and Central Africa (4). Nevertheless, these virulent genotype VI strains can also cause serious outbreaks in chickens (32, 33), for which they may be virulent, at least after circulation and adaptation (34, 35). At least two such outbreaks have been reported from South Africa (17) (subgenotype VIa) (Fig. 2B), where NDV surveillance is more intense. Although vaccination usually protects against clinical signs and mortality induced by virulent strains, not all farmers can afford the costs, and inappropriate vaccination may reduce efficacy. In addition, as a result of antigenic distances of these genotype VI strains from commonly used genotype I and II vaccines, vaccination may provide only incomplete protection (36, 37). Vaccines with at least a fusion protein sequence closer to the challenge strains would most probably provide a better immunity (38). Thus, the mingling of species in live-poultry markets increases the risk of transmission of genotype VI viruses to chickens, an additional risk factor for the poultry industry in Africa, already weakened by the circulation of other virulent NDV strains (4). Therefore, surveillance in domestic and wild Columbidae species is warranted to assess the risk associated with these virulent genotype VI strains.

Genotype XVIII.

One strain from a village weaver (Ploceus cucullatus, order Passeriformes) sampled in Côte d'Ivoire clustered within subgenotype XVIIIb together with chicken and duck strains from Côte d'Ivoire, Nigeria, and Togo (Fig. 2A). Genetic distances ranged from 0.4% for duck/Ivory Coast/CIV08-062/2006 to 3% for chicken/Nigeria/NIE11-1286/2011. Its cleavage site sequence, ¹¹²RRQKR↓F¹¹⁷, and the presence of a 6-nucleotide insertion in the intergenic region between the hemagglutinin-neuraminidase and polymerase genes (data not shown) were typical of virulent subgenotype XVIIIb isolates (4). This is compatible with the death of the bird, although another cause of death was not ruled out by a necropsy. This isolate represents so far the only case of the recently defined genotypes XIV, XVII, and XVIII in wild birds, except for two strains from a finch and a green wood hoopoe isolated from an import quarantine station in the United States (Fig. 2A) (39). In contrast to the spur-winged geese, village weavers were most probably infected by direct or indirect contact with sick poultry, since other cases of infection with genotype XVIIIb viruses were found in chickens in the same region in Côte d'Ivoire (4). Village weavers live in groups in urban or rural areas, feed on insects and seeds, and sometimes interact with free-roaming poultry during feeding. Similarly, cases of spillover of virulent NDV strains to wild birds were already described in China, where house sparrows (Passer domesticus) were infected with genotype VII strains similar to those causing outbreaks in poultry in the same area (40). The role of these synanthropic species in the epidemiology and spread of virulent NDV is unclear, since the virulence of the strains varies between species. However, morbidity and mortality of village weavers were likely to limit the spread of the virus. So far, in wild birds, virulent strains seem only enzootic in cormorants in North America (41) and in pigeons worldwide (42–44), but the two genotype XVIII cases detected in imported wild birds in the United States are puzzling. Therefore, monitoring infections in wild birds will be essential to better understand the host range of the virulent strains enzootic in West and Central Africa.

In our study, no class I strains were found in wild birds. Concern about these strains increased when some virulent viruses, derived from avirulent class I strains, caused outbreaks in Ireland in 1990 (45). They have never been found or systematically investigated in Africa, and there is no reason why they would not also occur on the African continent. Indeed, class I strains were identified in migratory birds in the United States (46) and Europe (47), and they could be introduced into wetlands of West Africa. Studies in live-bird markets in the United States showed that similar class I strains were found in wild waterfowl or shorebirds and poultry from live-bird markets (46). Ducks and chickens were also infected by class I strains in live-poultry markets in China and Hong Kong (48, 49). These two examples suggest that interactions with wild birds may result in virus transmission. Therefore, screening of poultry in markets in Africa could help to understand the epidemiology of class I in addition to wild bird sampling, which is much more difficult.

In summary, we have shown that several strains of avirulent and virulent NDV circulate in wild birds and pigeons in West Africa, in addition to the existing high genetic diversity of NDV strains found in poultry (4). Genotype VI strains were probably introduced on multiple separate occasions in Africa, and in Nigeria in particular. Typical avirulent wild-type strains were also found in wild birds. However, we did not find strong evidence that wild birds actively contribute to the spread of the enzootically circulating virulent genotype XIV, XVII, and XVIII strains. Further studies will be necessary to improve our understanding of the spread of virulent strains in West Africa and the role of wild birds.