Genetic diversity of avian leukosis virus subgroup J (ALV-J): toward a unified phylogenetic classification and nomenclature system.

Avian leukosis virus subgroup J (ALV-J) has infected a variety of birds, causing major economic losses in China. Understanding the comprehensive criteria of classification and nomenclature of ALV-J would be useful for the investigation of the viral evolution and also for the prevention and control of this infection. An in-depth analysis of the genetic diversity of ALV-J was performed in the present study. Four hundred and seventy-five sequences of the gp85 gene, including thirteen of avian endogenous retrovirus designated ev/J and 462 of ALV-J, were used in the phylogenetic and the evolutionary distance analysis for this classification. The study identified that the current ALV-J strains were divided into two first-order clades (Clades 1 and 2) and three second-order clades (Clades 1.1, 1.2 and 1.3). The current Chinese ALV-J strains are predominantly in Clade 1.3, and the Chinese and Egyptian chicken flocks have been facing the emerging Clade 2 viruses. This system pioneers the classification efforts for ALV-J, which uses Pilot tree for rapid classification of the new isolates and also the addition of possible new clades. The proposed unified classification system will facilitate future studies of ALV-J epidemiology and genetic evolution and of the comparison of sequences obtained across the world.

1. Introduction

Avian leukosis viruses (ALVs) that infect birds currently comprise eleven envelope subgroups designated A to K (Ma et al. 2020; Wang et al. 2020b). In the last thirty years, the importance of avian leukosis virus subgroup J (ALV-J) infections to avian health has been recognized. ALV-J is an oncogenic exogenous retrovirus and mainly induces neoplastic diseases and immunosuppression in infected birds (Payne, Gillespie, and Howes 1993; Zavala, Cheng, and Jackwood 2007). ALV-J has arisen through multiple recombination events between one or more endogenous and exogenous viruses (Bai, Payne, and Skinner 1995; Benson et al. 1998; Nair and Fadly 2013). ALV-J has the strongest pathogenicity and infectivity among all the subgroups, and it induces the most clinical chicken cases, estimated to account for more than ninety per cent of avian leukosis (AL) (Li et al. 2019). ALV-J was first isolated from commercial meat-type chickens in the United Kingdom in 1988 (Payne, Gillespie, and Howes 1992). Subsequently, ALV-J has caused severe economic loss to the poultry industry worldwide (Payne and Nair 2012). From 1999 to 2001, the virus infected the white-feathered broilers or parent stocks and then spread rapidly throughout China (Cui et al. 2003), and the host range of ALV-J gradually expanding to layer flocks (Cheng et al. 2010; Gao et al. 2012), Yellow-chickens (Sun and Cui 2007; Meng et al. 2018; Li et al. 2019), wild birds (Jiang et al. 2014; Shen et al. 2014; Zeng et al. 2014) and gamecock (Wang et al. 2018b). Generally, commercial chickens mainly include broilers and layers, and the broilers could be classified into White-chickens and Yellow-chickens in China. White-chickens are white-feathered broilers are mainly imported from the world-leading breeding companies from other countries. The Yellow-chickens refer to various indigenous chicken breeds and their hybrid offspring, which normally are colored-feathered chickens (not just yellow-feathered chickens). The broad circulation of ALV-J in poultry has led to significant genetic diversity of the virus and constant emergence of ALV-J. The outbreaks of ALV-J infection are still being reported in China since 2018 (Li et al. 2019; Zhou et al. 2019; Zhang et al. 2020a) even though eradication programs have been established in some big breeding companies (Wei and Cui 2015).

With the molecular methods of viral research, the knowledge of the ALV-J genetic makeup has been improved considerably. ALV belongs to the genus Alpharetrovirus and the family Retroviridae according to the International Committee on Taxonomy of Viruses (ICTV 2020). The proviral DNA arrangement of the genome of ALV-J is 5′ LTR-gag-pol-env-LTR 3′, which consists of approximately 7.2–7.8 kb nucleotides (nt), of which ALV-J env gene is distinct from the other ALV subgroups (Nair and Fadly 2013). The unique pathogenicity of ALV-J is closely related to its env gene, such that the ALV-J env gene could significantly increase the pathogenicity of the recombinant ALV-B (Li et al. 2020a). The envelope protein of ALV-J is encoded by its env gene and can be divided into gp85 and gp37 subunits. The gp85 protein, the viral surface knob-like structure, is closely associated with the process of viral binding and determines the viral envelope subgroup specificity of the ALV (Nair and Fadly 2013; Meng et al. 2016). The gp85 protein is the most variable of the structural proteins and exhibits high genetic diversity in the viral genome (Pan et al. 2012). Recently, it was reported that the bipartite sequence motif in the N (aa 38–131) and C (aa 159–283) termini of gp85 of ALV-J plays a crucial role in receptor binding and viral entry (Zhang et al. 2020b). The gp37 protein is responsible for mediating the fusion of viral protein with the host cell membrane and the C terminus of gp37 plays a vital role in ALV-J pathogenesis (Li et al. 2020b). At present, the need for ALV-J classification has resulted in a variety of methods for isolate differentiation. For example, a phylogenetic analysis based on the gp85 gene sequence found that viruses were related to the genetic background of infected chickens (Li et al. 2018). In the previous research of our research group, isolates from chickens of different genetic backgrounds were sorted into five clusters (Wang et al. 2018b), while the viruses were divided into two different clusters by the full-genome sequences analysis (Wang et al. 2020b). In order to better track the evolutionary history of the ALV-J, we now further propose a unified classification and nomenclature method. When using this approach, the names of the clades of higher order (closer to the tips), bear ancestral information from the lower clades they split from. No report similar to this method has been found. The gp85 gene sequence is usually used for the analysis of molecular epidemiology, and it is also the most detailed gene of ALV-J available in the GenBank database. A previous study showed that the gp85 gene of the env gene could better display the quasispecies diversity of ALV-J than the gp37 gene (Mao et al. 2013). In addition, the isolation and purification of the viruses are time consuming and expensive. The sequences of the gp85 gene of the current ALV-J strains are easier to obtain than their whole genome sequences. Other virus classification systems also use a single gene, such as the HA gene of H5N1 and the F gene of NDV, for classification (WHO/OIE/FAO-H5N1-EWG 2008 2014; Dimitrov et al. 2019). Therefore, it is currently appropriate to use the sequence analysis of the gp85 gene as an index to study the genetic diversity of this virus.

In this study, we employed phylogenetic methods to define the classification of ALV-J based on a total of 475 sequences of the gp85 gene. Our study provides insights into the classification and nomenclature of the current ALV-J strains.

2. Methods

2.1 Viruses sequences used in the study

Nucleotide sequences of the gp85 gene of 475 viruses were collected from our laboratory and from the GenBank National Center for Biotechnology (NCBI, http://www.ncbi.nlm.nih.gov/). One hundred and six sequences were generated (Supplementary Table S1), including eighty-eight sequences from our previous studies sampled during the years 2010–2018 (Lin et al. 2017; Wang et al. 2018a,b; Li et al. 2019; Wang et al. 2020b) and eighteen new sequences sampled during 2017–2020 (GenBank accession numbers: MW491254-MW491271), and 369 sequences with known sampling information, for example dates, locations and host were retrieved from the GenBank database, including 309 ALV-J strains from China sampled during 1999–2020 and forty-seven ALV-J strains from other countries sampled during 1988–2019, plus thirteen avian endogenous retrovirus designated ev/J (Supplementary Table S2). The analysis included nearly the full-length sequences of gp85 (e.g., >900 nt in length) to ensure robust statistical support. An initial alignment of the gp85 gene sequences dataset was aligned using MAFFT v7.149 (Katoh and Standley 2013) and then adjusted manually in BioEdit v7.2.5 (Hall 1999). The final alignment length was 924 nt. Viruses with 100 per cent sequence similarity were identified, giving a final alignment of 406 sequences. Viruses included in Supplementary Tables S1 and S2 that represent other viruses with 100 per cent sequence similarity sequences of the gp85 gene are listed in Supplementary Table S3. In order to further analyze the second-order clades of a dominant clade (Clade 1), a dataset of 387 sequences were further analyzed.

2.2 Phylogenetic analyses

To infer the phylogeny for the first-order clades and the second-order clades of the dominant clades, two phylogenetic trees from two datasets (n = 406 and n = 387) were respectively constructed using IQ-TREE v1.6.12 (Nguyen et al. 2015), and the best-fit model for two datasets were selected by following command: ./iqtree -s infile.phy -m MFP -nt AUTO, and then using the following command: ./iqtree -s infile.phy -m TVM+F+R6 -bb 1000 -nm 5000 -nt AUTO and ./iqtree -s infile.phy -m SYM+R6 -bb 1000 -nm 5000 -nt AUTO, respectively, for the two datasets. Meanwhile, two datasets used a maximum-likelihood (ML) approach in RAxML-HPC v.8 on XSEDE through the CIPRES Science Gateway v.3.3 (Miller, Pfeiffer, and Schwartz 2010) and the following command: raxmlHPC-HYBRID_8.2.12_expanse -f a -n tree -s infile.txt -m GTRGAMMAI -p 12345 -N 1000 -x 12345. All phylogenetic trees were visualized in FigTree v.1.4.3 (http://tree.bio.ed.ac.uk/software/figtree).

The final Clade 1 and Clade 2 were used to estimate the mean intra-clade distances (nucleotide distances between the first- and second-order clades) and the mean inter-clade distances. The estimates of the average evolutionary distances were inferred using MEGA7 (Kumar, Stecher, and Tamura 2016). The estimates method of average evolutionary distances is similar to the routine described by Dimitrov et al. (2019).

2.3 Pilot tree for rapid preliminary classification

After performing the analyses described above and identifying the main clades, two smaller (pilot) datasets of the gp85 gene sequences (n = 11 and n = 20) were parsed from the complete dataset used in the phylogenetic analyses using IQ-TREE, and the following command: ./iqtree -s infile.phy -m MFP -bb 1000 -nt AUTO. Two pilot datasets were phylogenetically analyzed with the ML method as described above to ensure that the topology inferred using the larger datasets (n = 406 and n = 387) would be maintained if fewer isolates were used.

2.4 Classification and nomenclature criteria of ALV-J

In all analyses, these criteria were used to initially assess the diversity of ALV-J and propose a consensus classification system: an ALV-J classification system that incorporates phylogenetic topology, genetic distances, branch support, and epidemiological independence was used; the names of the clades used Arabic numerals (e.g. 1, 2, 3…); and the ALV-J clades received a numerical-decimal address (Arabic numerals separated by periods) that starts with the Arabic numeral of the branch. Our classification criteria are in alignment with the World Health Organization/World Organization for Animal Health/Food and Agriculture Organization H5N1 Evolution Working Group (WHO/OIE/FAO-H5N1-EWG 2008, 2014) and the NDV Evolution Working Group (Dimitrov et al. 2019). Our classification and nomenclature criteria are summarized in Table 1.

Table 1.

Classification and nomenclature criteria for existing and new viruses.

No.	Description
1	A branch, closest to ADOL Line 0 (AF247392), designated as Clade 1.1. Subsequent clades numbered starting from Clade 1.2 (i.e., Clades 1.2–1.9).
2	Multiple sequence alignment using the MAFFT method. Assignment of different lengths of gp85 gene sequences with the reference virus ADOL Line 0 (AF247392) in batches.
3	Pilot tree can be used for assigning new viruses to existing clades.
4	Assignment of viruses into new clades is done based on near-complete gp85 gene from GGAGTTCAT to AGCAGGCGC (or longer) by phylogenetic analysis.
5	Assignment of viruses into new clades is done by utilizing only a complete dataset of gp85 gene sequences from the GenBank database.
6	New clades are created based on the phylogenetic tree topology (in a monophyletic groups) using the maximum likelihood method by IQ-TREE and RAxML software.
7	New clades are created only when four or more independent isolates, without a direct epidemiologic link, are available.
8	Estimates of evolutionary distances (mean nucleotide distance) between clades are inferred as the number of base substitutions per site from averaging over all sequence pairs between clades using MEGA 7/X software (or a comparable tool) and utilizing the maximum composite likelihood model with rate variation among sites that was modeled with a gamma distribution (shape parameter = 1).
9	Different first-order clades have an average distance per site above 10% (0.1). Different second-order clades have an average distance per site above 5% (0.05).
10	The ultrafast bootstrap value at the clade defining node is 70% or above.
11	All clades are identified using the Arabic numerals system to name. ALV-J clades receive a numerical-decimal address (Arabic numerals separated by periods) that starts with the Arabic numeral of the branch.
12	If a clade has unresolved topology, low branch support, or insufficient number of isolates, the viruses within this branch are assigned the name of the lower order until all criteria are fulfilled, regardless of the fact that the distance criterion may be correct.

3. Results

3.1 Phylogenetic analyses

The topologies of the phylogenetic trees, based on the sequences of gp85 gene (Fig. 1A and 1B), showed that all the ALV-J strains were divided into two first-order clades (Clades 1 and 2), respectively. Clade 1 is the main dominant epidemic clade and includes 455 viruses (13 ev/J and 442 ALV-J) from the UK, USA, China, Russia and Pakistan. Clade 2 is increasing and emerging and includes seventeen strains from the USA, China and Egypt (Table 2). In addition, although the genetic distance values of three recombinant strains JL09H01, JL09L01 and JL09DH02 in Jilin, China are >0.1 compared with other clades, they have not been reported in other provinces of China. Thus, no classification has been made (defined as ‘UNCL’) and these strains still need to be continuously monitored. Further phylogenetic analysis was performed to identify the second-order clades of Clade 1. Clade 1 was divided into three clades (Clades 1.1, 1.2 and 1.3) (Fig. 1C and 1D, Supplementary Tables S1 and S2). Clade 1.1, which has the widest range of sources, includes forty-four (44/156) strains from White-chickens, nineteen strains from layers, fifty-six strains from Yellow-chickens, one strain from the contaminated live vaccine, two strains from wild birds, five strains from gamecocks and plus thirteen ev/J, etc. Clade 1.2 primarily includes forty (40/88) strains from layers, twenty-eight strains from Yellow-chickens and twelve strains from wild birds. Clade 1.3 mainly includes forty-one (41/211) strains from White-chickens, twenty strains from layers and one hundred and twenty-one strains from Yellow-chickens. The current Chinese dominant ALV-J strains are in Clade 1.3. The strains from our laboratory were classified as clades 1.1 (14/106), 1.2 (14/106) and 1.3 (78/106).

Figure 1.

Phylogenetic analysis was performed based on the nucleotide sequences of the gp85 gene of ALV-J and ev/J. (A and B) Two radial phylogenetic trees were constructed based on the gp85 gene sequences (n = 406) of viruses representing ALV-J and ev/J to identify the first-order clades with IQ-TREE and RAxML software, respectively. (C and D) Two radial phylogenetic trees were constructed based on the gp85 gene sequences (n = 387) of viruses representing ALV-J and ev/J to identify the second-order clades of Clade 1 with IQ-TREE and RAxML software, respectively. The trees are drawn to scale, with branch lengths measured in the number of substitutions per site.

Table 2.

Details of strains obtained from Clade 2.

No.	Strains	Accession no.	Host	Location	Time
1	ADOL-7501	AY027920	W	USA	1997
2	1696	AF247384	W	USA	1997
3	6803	AF247388	W	USA	1997
4	6827	AF247389	W	USA	1997
5	AF88	AF247390	W	USA	1997
6	UD-J1	AF305091	N	USA	2000
7	SCSM00	KF796652	Y	Sichuan, China	2011-10
8	GDQJ-1	KU254611	N	Guangdong, China	2014
9	HB18XH01	MN735298	Y	Hebei, China	2018
10	SH18JY01	MN735306	Y	Shanghai, China	2018
11	SH18JY02	MN735307	Y	Shanghai, China	2018
12	QL1	MN496121	L	Egypt	2019-02
13	QL2	MN496122	L	Egypt	2018-11
14	QL3	MN496123	L	Egypt	2018-04
15	QL4	MN496124	L	Egypt	2019-05
16	QL5	MN496125	L	Egypt	2019-12
17	QL6	MN496126	L	Egypt	2018-08

W: White-chickens; L: layers; Y: Yellow-chickens; N: not available.

3.2 Evolutionary distances inferences

Distance analysis (Supplementary Tables S4 and S5) shows that the genetic distance between the two first-order clades was greater than ten per cent, and that among the three second-order clades was greater than five per cent. Within the second-order clade of Clade 1, the gp85 genes of Clade 1.1 are conservative (<4% variation) when compared with those of Clades 1.2 and 1.3. The main reason for this is that Clade 1.1 is in a cluster with the conservative ev/J. The estimations of evolutionary distances also confirmed that the classification of phylogenetic trees is appropriate.

3.3 Pilot tree and dataset

Here, we needed to find a rapid, preliminary method for clade identification of the new isolates, utilizing a phylogenetic tree with a lesser number of sequences that still provides topology consistent to the larger tree and clearly decreases the time for the tree construction. The eleven and twenty representative sequences of viruses were respectively used to perform ML phylogenetic reconstruction to distinguish the first-order clades and the second-order clades of Clade 1 (Fig. 2 and Table 3). These generated trees were consistent with the ML trees inferred using the larger gp85 gene dataset in Fig. 1. Thus, this pilot dataset can be utilized for rapid preliminary classification of the new isolates.

Figure 2.

Classification “pilot” tree. (A and C) Phylogenetic analysis are based on the gp85 genes of the selected viruses representing the first-order clade (n = 11) and the second-order clade (n = 20), respectively. (B and D) Use the newly isolated strains during 2017–2020 to test the topological structure of the tree. New isolates assigned in this current study are highlighted in red font. The trees are drawn to scale, with branch lengths measured in the number of substitutions per site.

Table 3.

Details of the representative viruses in the pilot tree of clades.

Number of viruses	Clades	Representative viruses (Accession no.)
11	1	ADOL Line 0 (AF247392), HPRS103 (Z46390), GX14YYA1 (MF461280), UD4 (AF307951), JL08CH3-1 (HQ634809), JS09GY2 (GU982307)
	2	AF88 (AF247390), QL1 (MN496121), SCSM00 (KF796652)
	UNCL	JL09H01 (HQ148554), JL09L01 (HQ148555)
20	1.1	ADOL Line 0 (AF247392), HPRS103 (Z46390), EAV-15I (AJ623289), 2921/00 (EF653908), DBYJ1101 (KC875863), GX14YYA1 (MF461280), UD4 (AF307951), BJ0301 (AY897230)
	1.2	JL08CH3-1 (HQ634809), JS09GY2 (GU982307), SVR807 (HM776937), GD1109 (JX254901), GX12NN03 (KT598489)
	1.3	GD14J2 (KU500032), GX12GL07 (KT598492), GDQY1201 (JX423792), 4817 (AF247385), GX14HG04 (KX058878), GX-J-4 (JQ246093), SD0001 (AY897223)

3.4 Nomenclature criteria

To ensure consistent naming, nomenclature criteria guidelines for naming clades are also outlined in Table 1. Most importantly, when using this approach, the names of the clades of higher order (closer to the tips) bear ancestral information from the lower clades they split off from, which helps to understand the epidemiology of ALV-J.

4. Discussion

ALV-J-related studies have been ongoing for more than thirty years; however, understanding the criteria for the classification and nomenclature of the viruses has remained poorly investigated. Although progress has been made in the control by the ALV eradication programs implemented since 2008 in China (Wei and Cui 2015), it still remains a threat, with quite a few cases in Chinese indigenous chickens reported every year. This comprehensive analysis of 475 sequences of the gp85 gene over the past three more decades will facilitate the understanding of the evolution of the virus, and the subsequent disease prevention and control.

In order to better track the origin of the virus, a more detailed classification and nomenclature method has been established here. Similar to our previous findings (Wang et al. 2018b, 2020b), viruses form independent clusters that are related to correspondingly host genetic backgrounds. In this study, we have improved the accuracy of the results in terms of classification methods and can now trace the source of the virus in terms of this nomenclature method. We first used the MAFFT software for sequence alignment, which aligned the different lengths of gp85 gene sequences with reference virus ADOL Line 0 (AF247392) in batches, removed the identical sequence using IQ-TREE and BioEdit software and then adopted the ML phylogenies generated by IQ-TREE and RAxML software, thereby greatly improving the stability of the evolutionary tree topologies. Additionally, some limitations of the methods are considered when interpreting our findings, but only 4.4 per cent (17/387) of all strains are inconsistent in classification by using the two methods. This analysis of the phylogeny did not change our conclusion. A recent study revealed that 18.1 per cent (3515/19,414) gene alignments from fifteen phylogenomic datasets yielded irreproducible phylogenies between two replicates using IQ-TREE, while 9.3 per cent (1813) genes yielded irreproducible phylogenies between two replicates using RAxML-NG (Shen et al. 2020). Compared with the results of IQ-TREE, the results of RAxML are more stable. Thus, the classification in this study mainly refers to the results of the RAxML. Selecting high-quality nucleotide sequences and sequence alignment using the MAFFT software and the ML phylogenies generated using multiple software will facilitate constructing a topologically stable evolutionary tree. The clades of the nomenclature suggested here are divided into three levels at most, including the first-order clades, the second-order clades and the third-order clades, so that it is easy to remember. ALV-J clades receive a numerical-decimal address (Arabic numerals separated by periods) that starts with the Arabic numeral of the branch.

In this study, the currently available ALV-J strains could be divided into two first-order clades (Clades 1 and 2) and three second-order clades (Clades 1.1, 1.2, 1.3) (Fig. 1). The evolutionary tree of Clade 1 made by our classification shows the spread of ALV-J from White-chickens to layers and Yellow-chickens. Clade 1.1 is a conservative cluster (Supplementary Tables S4 and S5). Clade 1.1 includes not only ALV-J isolates from various sources but also the 13 ev/J. Previous studies showed ALV-J is a multiple recombinant of the env sequence of ev/J and the exogenous avian leukosis virus (Bai, Payne, and Skinner 1995; Benson et al. 1998; Nair and Fadly 2013). Some of the ALV-J viruses in this cluster are closely related with the ev/J may be the relatively primitive viruses that have just been recombined the env sequence from ev/J. Also, it is worth noting that our isolate GX14YYA1 originated from the contaminated live commercial vaccine (Wang et al. 2018a) is in this cluster. Some of these strains are highly similar to the ALV-contaminated live vaccines isolate, and it is possible that they were transmitted through the contamination live vaccines and is contributing to the inability to eradicate ALV. They may contaminate the live vaccines produced by using the chicken embryo and/its cells. Clade 1.2 is mainly a branch of strains originating in layers. Clade 1.3 is dominated by the Chinese field strains originating in Yellow-chickens. Thirteen isolates from our laboratory were isolated by detecting the P27 antigen of ALV on the DF-1 cell cultures inoculated with the clinical tumor tissue samples at diagnosis and the sterile plasma samples of the birds at eradication testing during the years 2018–2020 (Supplementary Table S1). These were gathered in Clades 1.1 and 1.3, respectively, but the major clinical sample isolates (3/5) and the plasma sample isolates (7/8) were in Clade 1.3, revealing that incomplete eradication of ALV is one of the main reasons for the clinical AL cases. In addition, the ALV-J from the outbreaks in 2018–2019 in China had the closest genetic relationship with the isolate GD14J2 isolated from the White-chickens that have the origins from other countries (Ma et al. 2019; Zhang et al. 2020a). The Sichuan, Guangdong, Hebei and Shanghai isolates in Clade 2 obtained during the years 2011–2018 were found to be emerging, and which was most likely spread to China by the introduction of the White-chickens breeders from abroad (Wang et al. 2020a; Yu et al. 2014). Previously, few researchers paid attention to the pathogenicity and transmissibility of the six American strains (ADOL-7501, 1696, 6803, 6827, AF88 and UD-J1) in Clade 2. Now Clade 2 strains have been found in China and Egypt, where Clade 2 are gradually increasing and emerging, so it is necessary to continually monitor them (Table 2). Hence, four main clades (1.1, 1.2, 1.3 and 2) of ALV-J strains were revealed in China. There are three main causes for the increased genetic diversity of isolates from the commercial chicken flocks: the use of ALV-contaminated live vaccine/vaccines; the introduction of breeder chickens from abroad; and the incomplete eradication in the breeder flocks.

In conclusion, we first proposed the definition of the current ALV-J strains based on the gp85 gene that were divided into two first-order clades and three second-order clades. The current Chinese dominant ALV-J strains belong to Clade 1.3 and the Chinese and Egyptian chicken flocks have been facing the emerging Clade 2 viruses. This system includes comprehensive criteria for the classification of new ALV-J isolates and also the addition of the possible new clades. Our classification can play an irreplaceable role in tracking the evolution of viruses and in monitoring the spread of viruses currently.

Supplementary data

Supplementary data are available at Virus Evolution online.

Click here for additional data file.