Adaptive evolution of interferon induced protein with tetratricopeptide repeats 5(IFIT5) in birds


 Background : Previous studies have revealed that wild birds are reservoirs and mobile vectors of viruses, many of which cause illness and mortality in domestic bird and humans. In birds, the invasion of viruses will quickly trigger the innate immune mechanism induced by interferon (IFN). As IFN-stimulated genes (ISGs), the IFIT gene family plays an important role in innate immunity. However, only IFIT5 of the IFIT gene family exists in birds, and the direction and strength of selection acting on IFIT5 are largely unknown. Results : Here, we studied the selection on IFIT5 based on the coding sequence (CDS) data of 20 birds. We identified 12 persistent positive selection sites (PSS), other sites suffered purifying selection and neutral selection; probably due to functional constraints. We also found humans have only 3PSS (189,197and 295), likely due to having more IFIT gene family member that can cooperate to resist virus invasion. The 12 PSS located in the closed clamp structure of the IFIT5 protein, except for position 45. In particular, 3 PSS (335, 342 and 367) were located in the TPR domain, which implied their important roles in virus recognition. We only found 2 episodic PSS (30,332) in Passeriformes, indicating episodic selection pressure in Passeriformes lineage. The positive selection of IFIT5 might provide a theoretical basis for the pathogen-host interaction in birds. Conclusions : We found that the diversity of IFIT5 domains in birds, and that the PSS of IFIT5 is the joint influence of functional domain conservation and the pressure of virus evolution.We speculated that persistent PSS may affect the antiviral function of IFIT5, especially in the region of closed clamp structure. These results lay a theoretical foundation for the further study of the antiviral immune mechanism of IFIT5 in birds.


Background
The birds are a part of the subphylum Vertebrate, with more than 10,000 species (www.iucnredlist.org).
Birds are warm-blooded animals. The body temperature can be as high as 40℃; they are oviparous animals with amniotic membranes outside their embryos; most birds can fly with evolutionary wings on their forelimbs (Brusatte et al. 2015). Previous studies implied that wild birds are reservoirs and mobile vectors of pathogens because most birds have migratory characteristics (Jeon et al. 2008). Studies have found that birds can be infected or carry a variety of pathogens, including avian influenza, avian cholera, chicken anemia virus, Newcastle disease, West Nile fever and other viral and bacterial diseases (Benskin et al. 2009;Botzler 1991;Bush and Clayton 2018;Chatziprodromidou et al. 2018;Fatoba and Adeleke 2019;LaDeau et al. 2007;Rahman et al. 2018).
These pathogens bring potential threats to wild birds, domestic animals and humans. In particular, the H7N9 subtype of avian influenza not only causes heavy damage to the economy of the poultry industry, but also causes human casualties (Chatziprodromidou et al. 2018;Li et al. 2019;Wang et al. 2017). However, when birds are invaded by the virus, the innate immune system will respond immediately, and the relevant immune genes will be quickly expressed to resist the invasion of the virus.
Interferon-stimulated genes (ISGs) play a main role in the innate immunity of birds (Gimeno and Schat 2018). ISGs has a large number of members, and their structure and cell location are also different, which indicates that different members play an antiviral role by different mechanisms (Crosse et al. 2018;Santhakumar et al. 2017;Schoggins 2018). Interferon-induced proteins with tetratricopeptide repeats (IFIT) gene family is an important member of the ISGs gene family, which plays an indispensable role in inhibiting viral transcription and replication (Zhou et al. 2013).
The IFIT family is interferon-stimulated genes induced by interferon, which participated in the innate immunity of birds (Vladimer et al. 2014;Zhang et al. 2019). The IFIT gene family only exists in mammals, birds, fish and other higher animals. The IFIT protein family has four members: IFIT1 (ISG56), IFIT2 (ISG54), IFIT3 (ISG60) and andIFIT5 (ISG58) (Fensterl and Sen 2011). The IFIT gene family has a similar gene structure, which is composed of two exon and one intron. Exon 1 has 5'-UTR region and start codon and exon 2 is the coding region and 3'UTR genetic information. The IFIT family proteins are mainly consisted of Tetratricopeptide repeats (TPR) domains, TPR domain can mediate protein-protein interaction (Feng et al. 2013;Fensterl and Sen 2011). The study found that IFIT1 and IFIT5 proteins have a closed clamp structure with positive charge inside and negative charge outside (Fensterl and Sen 2015).
This structure can specifically bind to 5'-pppRNA, thus affecting the translation of viral RNA (Abbas et al. 2013). IFIT1 is required to combine with IFIT2 and IFIT3 to form a complex to stabilize the conformation of IFIT1 protein and play to its antiviral function (Mears and Sweeney 2018). However, IFIT5 can be combined with viral 5'-pppRNA alone to play an antiviral role (Abbas et al. 2013).
Interestingly, IFIT5 is only member of the IFIT gene family that exists in birds (Fensterl and Sen 2011).
IFIT5 plays a significant role in defense the invasion of viruses and protecting the normal physiological environment of birds.
The study found (Johnson et al. 2018)that the mutation of IFIT gene may change the structure of the protein, which affects the stability and antiviral function of the protein. (Feng et al. 2013)found that mutations at some specific sites can weaken the binding ability of IFIT5 protein to RNA. In particular, the mutation of K415/R384E site will seriously damage the ability of IFIT5 protein to inhibit the reproduction of the virus. The mans that the mutation of a single site of IFIT5 may also affect the 3 antiviral function of the protein. At the molecular level, the pressure of evolution can act on changes amino acid sites (Forslund et al. 2019). Based on these previous studies, we can infer that there is a co-evolution "arms race" history between IFIT5 and viruses (Daugherty and Malik 2012). Viruses exert biological pressure on the evolution of IFIT5 proteins, which make them have higher antiviral efficiency.
Despite the increasing interest in IFIT5 as a crucial member of the IFIT gene family, nothing is known so far about the adaptive evolution of molecular level in birds. Based on NCBI public database(https://www.ncbi.nlm.nih.gov/). We downloaded the IFIT5 mRNA sequences of 18 species of birds and human, and then sequence the CDS of Bashang long-tailed chicken and White Leghorns chicken. Specifically, we studied the phylogenetic tree, the domain architecture and diversity, the strength and direction of selection, focus on the persistent positive selection sites found in the protein sequence. Finally, we discussed the position of the positive selection site in the 3-D structure of Gallus gallus IFIT5 protein and its possible effect on the antiviral function of the protein.

Domain architecture and diversity
In this study, the bird species IFIT5 is a protein structure with a typical tetratricopeptide repeats. The TPR motif is a degraded sequence of about 34 amino acids, loosely located around the common residue-WLG-Y-A-F-A-P-,which mediates a variety of protein-protein interactions(D'Andrea and Regan 2003).
There are some differences between the number of TPR and species by the SMART predicted. The IFIT5 of most birds, such as chicken, has six TPR domains, while the IFIT5 of duck and Apteryx rowi (ancestral lineage) has five TPR domains. However, results predicted by humans IFIT5 are only 5 TPR domains, which is different from previous studies. After correction, human ifit5 has 9 TPR domains (Feng et al. 2013). The number of TPR in diverse species may affect the antiviral function of IFIT5.
The results show that the protein sequence of IFIT5 in birds, in addition to the typical TPR domain, also has the low-complexity region (LCR) region, proteinphosphatase2A catalytic subunit (PP2Ac) and coiled-coil (CC) region. PP2Ac can affect the IFN α signal pathway. Thus affecting the early immune response (Shanker et al. 2013). Analysis shows that both Anas platyrhynchos and Nothoprocta_perdicaria have PP2Ac.Coiled-coil is a single and regular domain, which plays an important role in the structure and function of proteins; it involves biological processes such as molecular recognition, signal transduction, and ion channels (Burkhard et al. 2001).The results show that only Taeniopygia guttata has coiled-coil. LCR domains have high genetic abundance and can be expanded in a relatively short time by replication sliding, so they can greatly increase the protein sequence space (Coletta et al. 2010;Golding 1999).Interestingly, there are differences in the number of LCR among different species. There are three Apteryx rowi, two in Gallus gallus and only one in Homo sapiens.
Bashang long-tailed chicken is a characteristic breed in Hebei Province, China, with excellent traits such as cold tolerance, rough feeding tolerance, high meat quality, strong immunity. White Leghorns chicken is a typical commercial laying breed, which has the characteristics of high egg production and low feed consumption. We obtained the IFIT5 coding sequence (CDS) of IFIT5 of these two breeds and compared them with the orthologs of duck and human (Supplementary table S2). The Bashang longtailed chicken CDS sequences showed 79.05% and 59.89 % sequence identity to their duck and human orthologues；White Leghorns chicken results were 79.19% (duck)and 59.82% (human).The adopted Bashang long-tailed chicken IFIT5 protein sequences showed less similarity with 67.78%(duck)and 48.53% (human) amino acid sequence identity；this result also belongs to White Leghorns chicken. Compared to the White Leghorns chicken, the CDS similarity of the Bashang long-tailed chicken is lower, which may indicate that it has a higher polymorphism, so as to offer raw materials for evolutionary selection.

Evidence for persistent positive selection
We performed site model analyses to detect signatures of persistent positive selection in specific bird sites within IFIT5 evolution. According to model M2a the large majority of codons evolved under purifying selection (43 %, ω=0.19), 44 % of sites evolved neutrally (ω=1) and 13% of all sites were detected to evolve under positive selection (ω=2.73; Table 2). When models that estimate ω with a beta distribution, the M8 model predicted that 18% of the sites were in positive selection (ω = 2.35; Table 2).
In order to raise the reliability of PSS, we used M1avs.M2a M7vs.M8 pairwise model for comparison (Kosakovsky Pond and Frost 2005). M2a and M8 models are identified as w> 1, this site is considered to be a positive selection site (displayed in bold). We found that the TPR region is still relatively conservative, and only three PSSs (position 335, 342, and 367) are predicted.12 specific codon sites were found to be positive selected sites (PSS) by two pairs model analyses and were thus elected as putatively PSS in birds. Amino acid character alterations were present in all 12 PSS including non-polar and polar as well as basic polar and acidic polar side chain polarities (Table 3). By the amino acid conservation scores of the ConSurf analyses, which indicated all PSS to be variable (scores<5 , Table 3)

Evidence for episodic positive selection
We performed branch-site analyses to detect signatures of episodic positive selection in specific birds line-ages within IFIT5 evolution. We used ModelA vs ModelA null for testing, which is an excellent detection models pair (Zhang et al. 2005). We conducted an episodic positive selection test for the ancestral lineage and higher taxonomic branches, including the ancestral branch, Galliformes, Passeriformes, Psittaciformes and Falconiformes. Analyses indicated that most bird branches don't predict the PSS or the PSS isn't significant. Only two PSS (30D, 332K) were detected in the Passeriformes branch (Table 4, Fig. 1). The outcome suggested that Passeriformes is under significant episodic selection pressure. Both position 30 and 332 aren't in the TPR domain of the IFIT5 protein, these two sites are relatively conservative sites (Conservation score 6) by Consurf predicted.

Predicted molecule structure and location of PSS according to Gallus gallus IFIT5
Analyses of surface accessibility confirmed that all PSS, except two (aa position 45and 335), should prepossess exposed positions in the three-dimensional molecule structure (Table 3). The results showed that only 3PSS (335, 342and 367) were located in the TPR domain, and the remaining 9 PSSs weren't located in the domain (Table 3).However, the results shows on Gallus gallus 3D protein model that most of the PSS are situated in the protein closed clamp structure (Fig. 2).The closed clamp structure can specifically bind to the virus 5′-pppRNA and inhibit the replication of RNA virus (Abbas et al. 2013;Feng et al. 2013). These imply that PSS is on an important region of the protein, especially the three PSSs located in the TPR domain.
The second region is within the TPR5 and there are three PSS (335,342and367). The last region is upstream of TPR6, with 2 PSS (404 and 409). In addition, position45 is situated in the upstream region of TPR1 and position295 is located in the downstream area of TPR4.Position295 was also identified as PSS in birds and human.

Discussion
The immune function of organisms can be consisted into three levels: physical defense, innate immunity and adaptive immunity. Innate immunity is an inherent immunity of organisms, which can quickly identify invasive pathogens and trigger the initial immune response, which is an important line 5 of defense of the immune system (Liu et al. 2018;Negishi et al. 2018;Zhou et al. 2013). Among them, ISGs plays an indispensable role in the innate immune system. There're differences in the antiviral function of different ISGs (Villalón-Letelier et al. 2017), the IFITM gene family is mainly resistant to the invasion of the virus, and the Viperin family can effectively inhibit the release of the virus.
Interestingly, the IFIT gene family is mainly targeted at the replication and transcription stage of the virus; it's an essential member of the ISGs (Fensterl and Sen 2015). Studies have shown that the IFIT gene family is resistant to a variety of pathogens; chlamydia, bacterial lipopolysaccharide and viruses can induce the expression of IFIT (Diamond and Farzan 2013). It is found that the members of the IFIT gene family are relatively conservative, and the structure is mostly composed of two exons and one intron (Vladimer et al. 2014). IFIT protein family is a cytoplasmic protein and has the characteristics of multiple TPR domains. There are differences in the number of TPR owned by members of the IFIT family: IFIT1 has 6 TPR domains, and IFIT5 has 9 TPR domain (Cerveny et al. 2013;Fensterl and Sen 2015). IFIT gene family exists in birds, fish, mammals and other higher organisms. Interestingly, there are different in all IFIT gene family members of the species (Fensterl and Sen 2011). Humans have all IFIT gene family members, IFIT3 does not exist in gibbons, IFIT5 does not exist in mice, and the single member of birds is IFIT. As the only family member of birds, IFIT5 plays an important role in the process of antivirus (Zhou et al. 2013). As an immune gene, IFIT5 bears the biotic pressure to resist the invasion of viruses, so it is suitable for adaptive selection correlation analysis. The analysis of adaptive selection of IFIT5 is helpful to provide the theoretical basis for disease resistance breeding of birds.
The IFIT5 of birds is similar to that of human IFIT5 gene, and both of them are conserved in gene and domain structure, and both have antiviral function (Feng et al. 2013;Fensterl and Sen 2011). In the bird species studied, the coding sequences of IFIT5 exist in exon 2 and have typical TPR domains, which are consistent with the structure of human genes. However, long-term evolutionary pressure has resulted in differences in the length of IFIT5 CDS sequences in different birds. In addition, the species, number and location of domains are also different among different species. Most birds IFIT5 proteins have 6 TPR domains, while Apteryx rowi have only 5 TPR, and 9 TPR in humans. However, Apteryx rowi have three LCR domains, while chickens have two LCR, and only one in humans. The LCR domain has a high genetic abundance, which can increase the protein sequence space and evolve a specific functional domain (Coletta et al. 2010). Previous studies have shown that the number and type of domains have a certain effect on the function of the protein (Forslund et al. 2019). Compared with chickens and humans, Apteryx rowi has more LCR and less TPR. We speculate that these may be the result of large-scale evolution of the structural domain of IFIT5 lineal ortholog genes.
In this study, we cloned the complete CDS region of IFIT5 of Bashang long-tailed (BS) chicken and White Leghorns (LH) chicken. Through sequencing, we obtained the nucleic acid sequence, and then deduced the amino acid sequence. Therefore, we discuss the diversity of sequences. The sequence alignment of these two breeds and Gallus gallus showed that the sequence homology was as high as 99%, indicated that the cloning result was correct; the homology of BS chicken sequence was lower, indicated that BS chicken had higher sequence diversity. BS chicken and LH chicken were compared with ducks and humans respectively. The results showed that chicken IFIT5 and duck IFIT5 had high homology compared with humans, which were consistent with the results of the phylogenetic tree.
Interestingly, the alignment results showed that the protein sequence homology was the same ones, while the CDS nucleic acid sequence homology of BS chicken was lower. This shows that Bashang long-tailed chicken, as a chicken breed with strong immunity, has high polymorphism in CDS sequence, which provides raw materials for the molecular evolution of IFIT5.
Previous studies have shown that immune-related genes are often subjected to the pressure of evolutionary selection because they need to resist the invasion of pathogens (Manry et al. 2011).In birds, we found that most of the IFIT5 CDS sites were under purifying selection and neutral selection pressure, and a small number of sites were subjected to positive selection pressure (Table2).The implies that the IFIT5 gene of birds is relatively conservative. We compared paired models of M1avs.M2a and M7vs.M8, and the common positive selection sites. The results are show that birds have more PSS than humans. 12 PSS were predicted and only 3 sites (189,197and 295) were discovered in the human protein sequence. We speculated that compared with only IFIT5 of birds, humans have more members of IFIT gene family, which can cooperate to resist the invasion of viruses.
Therefore, the function of human IFIT5 gene tends to be stable and the selection pressure is low.
In this study,12 PSS were found, of which 3 PSS ( Passeriformes are the most complex birds with complex living habits and are subject to strong pressure of evolutionary selection (Knight et al. 2018). In this study, the number of passerine species is large, and the sequences of other bird families are few; it may cause other bird families to fail to detect episodic PSS (Roiz et al. 2019). We assume that these two sites may be potential functional sites of passerine IFIT5 protein.
We projected 12 persistent PSS onto the predicted 3D protein model, revealed that most of the sites were located in the closed clamp structure of the protein, except for site45.Previous study found that multiple TPR domains of IFIT5 protein form a right super helix, which forms a closed clamp structure, most of which can be used to bind ligands (Fensterl and Sen 2011).Interestingly, the closed clamp structure has a positive charge on the inside and a negative charge on the outside. Through the closed clamp structure, IFIT5 can specifically bind to the negatively charged virus 5'-pppRNA and change the virus RNA sequence, thus inhibiting the replication of virus RNA (Abbas et al. 2013).Previous studies have shown that the extremely conservative sites of the receptor recognize the invariant residues of the ligand, while the ligand residues recognized by the positive selection site are variable (Daugherty and Malik 2012;Schuster 2018). This indicates that the immune function of IFIT5 protein is affected by both conserved domain and virus selection pressure. 11 sites of the closed clamp structure are potentially evolutionary functional sites. The results showed that 10 amino acids were exposed and 2 were hidden (Table 3). It is predicted by ConSurf that the score of 7 family can change the structure of the encoded protein and significantly affect the stability of the protein (Johnson et al. 2018). These PSS play an important role in the function of the protein and affect the virus recognition of the protein.
In this study, we found that 12 persistent PSS projected onto the protein 3D model were mainly distributed in three aggregation regions, except for two free sites (45,295).We found that one region was located between TPR2 and TPR3, the second region was within TPR5, and the last region was located in the upstream region of TPR6.The TPR domain can mediate a variety of protein-protein interactions and plays an indispensable role in the immune function of IFIT5(Perez-Riba and Itzhaki 2019). The study found that PSS aggregation regions can predict the regions of host-virus interaction (Daugherty and Malik 2012). In summary, the results of this study are similar to the conclusions of Daugherty. Previous studies have shown that hosts can evolve resistance to viral infection through selective mutations of amino acids in the binding region of receptor proteins (Suárez-Díaz 2016). We detected that the persistent PSS of the three aggregation regions may have a similar evolutionary direction, thus enhancing the disease resistance of the host.
The study found (Zhang et al. 2019) that duck IFIT5 can effectively resist duck hepatitis A virus type3 (DHAV-3) and play an important role in anti-DHAV-3 infection. (Vanderven et al. 2012) found that the expression of IFIT5 protein increased significantly in infected with highly pathogenic avian influenza virus A in ducks. Interestingly, RNA virus can prevent the normal recognition of IFIT protein by adding 5'-cap structure and 2-O methylation, thus avoiding the elimination of IFIT gene family (Daffis et al. 2010;Szretter et al. 2012). The "arms race" between hosts and viruses exists in the history of biological evolution for tens of thousands of years (Suárez-Díaz 2016). However, changes in amino acid sites may influence the structure and function of the protein (Shastry 2002;Shatoff and Bundschuh 2020). Therefore, the adaptive evolution of sites reflects the direction and degree of evolution of host immune proteins. In the co-evolution of birds and their viruses, it could be speculated that the antiviral effect of IFIT5 may be enhanced by weakening the escape mechanism of the virus by inhibiting the exact ligand binding site. Therefore, the PSS increase the potential of birds to face future environmental changes and promote the viability of the population.

Conclusions
In summary, our results suggest the diversity of IFIT5 domains in birds, and that the PSS of IFIT5 is the joint influence of functional domain conservation and the pressure of virus evolution. We speculated that persistent PSS may affect the normal immune function of IFIT5, especially in the region of closed clamp structure. These results lay a theoretical foundation for the further study of the antiviral immune mechanism of IFIT5 in birds.

Sample collections and genomic extraction
The Bashang long-tail chicken sample was presented by Dr. Xiaohui ,Liu from Hebei Agricultural University(already graduated). White Leghorns (LH) chickens are stored in the laboratory. Only the experimental animals(chickens) were blood tooken from the veins under the wings , without slaughter.
Blood samples were taken from a wing vein of individuals for genomic DNA extraction from erythrocytes. Then took back to the laboratory after heparin anticoagulation (stored at 4℃). Genomic extraction was performed using of standard phenol-chloroform method (Wang and Yu 2014).

DNA amplification and sequencing
Most of the CDS of the IFIT5 is located in exon 2, and exon 1 has only the start codon. Exon2 of IFIT5 were divided into two parts to design primer due to long amplification products. Primers were referred from previous study (Li et al. 2017) (Supplementary table S1).
PCR was performed in a total of 25μL of reaction volume containing 1.0μl (100 ng/μl) of template, 1μL of each primer (10pmol/μl), 12.5μL2×Es-Tap MasterMix (Dye)and 9.5μL ddH2O. PCR conditions were as follows: initial denaturation at 94°C for 4 min followed by 30 cycles of denaturation at 94°C for 30s, annealing between at 57°Cfor 30s and elongation at 72°C for 1min depending on fragment length and a final elongation step by72°Cfor another5 min. The PCR products were characterized in 1.5% agarose gel electrophoresis and sequence by Chain Termination Method in BeiJing SaiBaiShi biological technology Company Limited. Chicken sequences have been filed in GenBank under the following accession numbers (MW045208, MW045209).

Phylogenetic analyses
The IFIT5 sequences of 20 other bird species, and human as outgroup were retrieved from GenBank. Available IFIT5 mRNA sequences of other bird species were contained in our analyses. The GenBank accession numbers are listed in Table 1. The codon sequences were aligned using MUSCLE implemented in MEGAX. Then, the maximum likelihood (ML) trees were constructed using the program MEGA X, based on the amino acid -codon sequences (Yang 1994).The substitution model for each dataset was aligned by Kimuras two-parameter evolutional model(Kimura 1968).Bootstrap percent probabilities were based on 500 replicates. The scale bar represents a genetic distance of 0.02 (Kumar et al. 2018).

Selection detection
The nonsynonymous (dN)/synonymous(dS) substitution rate ratio is usually expressed as ω (Ellegren 2008). The ω (dN/dS) value is an extremely important parameter for evaluating the selection detection.
There are three cases of ω value: the first one, ω = 1, the site is subject to neutral selection; the second, ω> 1, the site is subject to positive selection; the third, 0 <ω <1, the site under negative selection, also known as purifying selection (Yang et al. 2000). Neutral sites do not affect biological functional, however, positive selection sites will be fixed in the population, and negative selection sites will be gradually eliminated.
The CODEML program used to use PAMLX, which estimates ω among sites applying different models of codon evolution (neutral, purifying, or positive) (Xu and Yang 2013). However, PAMLX have a few limitations. We utilized EasyCodeML software analysis codon evolution. EasyCodeML that provides an user-friendly graphical interface and simple operation for using CodeML(Gao et al. 2019).

Site models
This model assumes that different branches of the phylogenetic tree are under the same selection pressure, considering only the ω values of different amino acid sites (Yang et al. 2000). The positively selected sites were identified under Bayes empirical Bayes (BEB) with a posterior probability of P > 95% . Model M0(one-ratio) and M3 (discrete) assume a constant ω across all sites. Model M1a (nearly neutral) and M7 (beta) assume sites under neutral or negative selection. Models M2a (positive selection), and M8 (beta and ω > 1) assume sites under positive selection. Models M8a (beta and ω = 1) is similar to M8 except that the ω value of the site is fixed at 1. The sites were by mutual comparison of paired models, which were confirmed to be positively selected sites (Kosakovsky Pond and Frost 2005).

Branch-site models
In this model, supposed the ω value of the sites is changed and different branches of the phylogenetic tree are also changed. These codon substitution models are: Model A (model=2 NS sites=2), Model B (model=2 NS sites=3), Model C (model=3 NS sites=2), Model D(model=3 NS sites=3).These models are used to identify significant episodic selections in specific branche ).s.The Comparison of Model A(positive selection along specified branches)and Model A null(neutral evolution and negative selection)can increase the accuracy of detecting positive selection sites (Zhang et al. 2005).To test for episodic positive selection Branch-Site Model A and Model A null implemented in the EasyCodeML software were used.

Structure analyses of bird IFI5
The ConSurf (Haim et al. 2016)was used to analysis the conserved of persistent positive selection sites(https://consurf.tau.ac.il/). The Phyre2 (Kelley et al. 2015)was used to predict the surface accessibility (buried or exposed) and the secondary structure (helix, coil or strand) for each sites, which we analyzed the potential functionality of sites under positive selected. To identify the domain of positively selected sites of IFIT5, we analyze the amino acid sequences using SMART (http://smart.emblheidelberg.de/) online analysis platform(Letunic and Bork 2017). The SWISS-MODEL(https://swissmodel.expasy.org) was used to predict the tertiary structure of the protein, and base on the crystallised Gallus gallus IFIT5 structure (Marco et al. 2014). Edited the tertiary structure and  a 2Δ(lnL) twice the difference in 1nL between the two models compared; b p0 proportion of sites were ω<1, p1 proportion of sites were ω=1, p2 proportion of sites were ω>1;