Ancestral Sequence Reconstruction for Viral Evolution: Methods, Applications, and Insights in Veterinary Virology

Introduction

Ancestral sequence reconstruction (ASR) is a computational approach that infers the genetic sequences of extinct or ancestral viruses from contemporary sequence data using phylogenetic models [1, 2]. In veterinary virology, ASR provides a powerful framework to investigate the evolutionary pathways that underlie host range expansion, receptor binding specificity, antigenic drift, and the emergence of zoonotic pathogens [3, 4]. By reconstructing ancestral viral proteins, researchers can experimentally characterize the biochemical and biophysical properties of ancient molecules, thereby identifying key mutations that enabled cross-species transmission or immune evasion [4, 5].

The core principle of ASR relies on a well-resolved phylogenetic tree and a model of nucleotide or amino acid substitution [1, 6]. Maximum likelihood and Bayesian inference are the predominant statistical frameworks used to estimate ancestral states at internal nodes of the tree [2, 7]. These methods account for the stochastic nature of molecular evolution and provide posterior probabilities for each inferred residue [3, 6]. The accuracy of ASR depends on the quality of the sequence alignment, the appropriateness of the substitution model, and the taxonomic sampling density [1, 4].

In veterinary contexts, ASR has been applied to diverse viral families, including orthomyxoviruses (influenza A virus), coronaviruses (e.g., SARS-CoV-2 and related bat coronaviruses), and retroviruses (e.g., feline immunodeficiency virus, FIV) [2, 3, 4]. The technique enables the identification of ancestral glycoprotein conformations, the reconstruction of ancestral receptor-binding domains, and the prediction of historical host tropism [3, 4]. This article provides a detailed technical review of ASR methodology, its applications in veterinary viral evolution, and its limitations, with a focus on computational and biophysical principles.

Methodology of Ancestral Sequence Reconstruction

Phylogenetic Inference and Substitution Models

ASR begins with the construction of a phylogenetic tree from a multiple sequence alignment of extant viral sequences [1, 6]. The tree topology and branch lengths are estimated using maximum likelihood or Bayesian methods under an explicit model of sequence evolution [2, 7]. Commonly used substitution models for nucleotide sequences include the general time-reversible (GTR) model with gamma-distributed rate heterogeneity [3]. For amino acid sequences, empirical matrices such as JTT, WAG, or LG are employed, often with a gamma correction for among-site rate variation [4, 5].

The choice of substitution model significantly influences the accuracy of ancestral state estimates [1, 6]. Model selection is typically performed using information criteria such as the Akaike information criterion (AIC) or the Bayesian information criterion (BIC) [2, 7]. In viral evolution, where substitution rates are high and selection pressures are strong, codon-based models that separate synonymous and nonsynonymous substitution rates may provide more biologically realistic reconstructions [3, 4].

Ancestral State Estimation

Given a phylogenetic tree and a substitution model, ancestral sequences are inferred at each internal node using either marginal or joint reconstruction [1, 6]. Marginal reconstruction calculates the posterior probability of each possible residue at a given node, independent of other nodes, while joint reconstruction finds the combination of residues across all nodes that maximizes the overall likelihood [2, 7]. Maximum likelihood methods use the pruning algorithm to compute the likelihood of ancestral states, whereas Bayesian methods incorporate prior distributions on model parameters and tree topology [3, 5].

The output of ASR is a set of inferred ancestral sequences, each associated with a posterior probability for every site [1, 4]. Sites with low posterior probabilities (e.g., below 0.7) are considered unreliable and are often excluded from downstream experimental characterization [2, 6]. For veterinary applications, the reconstruction of full-length glycoprotein genes (e.g., hemagglutinin, spike protein, envelope) is common, as these proteins mediate host cell entry and are primary targets of neutralizing antibodies [3, 4, 7].

Experimental Validation of Ancestral Proteins

A critical step in ASR studies is the synthesis and functional testing of inferred ancestral proteins [4, 5]. Synthetic genes encoding the ancestral sequence are cloned into expression vectors and used to produce recombinant proteins or pseudotyped viruses [3, 7]. Functional assays include receptor binding affinity measurements (e.g., surface plasmon resonance), cell entry assays, and neutralization assays with sera or monoclonal antibodies [4, 5]. This experimental validation confirms that the reconstructed ancestor is biologically plausible and provides mechanistic insights into evolutionary transitions [1, 4].

flowchart TD
    A[Collect extant viral sequences], > B[Multiple sequence alignment]
    B, > C[Phylogenetic tree inference]
    C, > D[Select substitution model]
    D, > E[Ancestral state estimation (ML or Bayesian)]
    E, > F[Reconstructed ancestral sequences]
    F, > G[Synthesize ancestral genes]
    G, > H[Express recombinant proteins]
    H, > I[Functional assays: receptor binding, entry, neutralization]
    I, > J[Interpret evolutionary adaptations]

Applications in Veterinary Virology

Host Adaptation and Receptor Binding in Influenza A Virus

Influenza A virus (IAV) circulates in avian and mammalian hosts, with occasional spillover into swine and humans [4]. ASR has been used to pinpoint the molecular adaptations that enabled avian IAV to transmit in pigs, a key intermediate host for pandemic strains [4]. Su et al. reconstructed ancestral hemagglutinin (HA) sequences at nodes preceding the emergence of swine-adapted lineages [4]. Functional testing of these ancestral HAs revealed that a single amino acid substitution in the receptor-binding site (e.g., at position 190 or 225) shifted binding preference from avian-type alpha-2,3-linked sialic acids to mammalian-type alpha-2,6-linked sialic acids [4]. This study demonstrated that ASR can identify critical residues that govern host tropism, informing surveillance efforts for emerging zoonotic IAV strains [4].

The same approach has been applied to the neuraminidase (NA) protein, where ancestral reconstructions identified mutations that conferred resistance to antiviral drugs such as oseltamivir [4]. In veterinary medicine, these findings are relevant for managing IAV outbreaks in swine and poultry, as drug resistance can compromise therapeutic options [4].

Coronavirus Spike Protein Evolution and Zoonotic Potential

Coronaviruses, including SARS-CoV-2, have emerged from animal reservoirs, with bats and pangolins serving as key hosts [3]. Brintnell et al. used ASR to reconstruct the ancestral spike protein of SARS-CoV-2 and related bat coronaviruses [3]. The reconstructed ancestral spike exhibited a latent capacity to bind the human angiotensin-converting enzyme 2 (ACE2) receptor, even though contemporary bat coronaviruses showed weak or no binding [3]. This finding suggested that the progenitor of SARS-CoV-2 possessed a pre-existing ability to interact with human ACE2, which was subsequently refined through a small number of mutations [3]. For veterinary virology, ASR of coronavirus spike proteins can help assess the zoonotic risk of circulating animal coronaviruses, such as those in cats, dogs, and livestock [3].

Lentiviral Evolution and Immune Evasion

Lentiviruses, such as FIV and equine infectious anemia virus (EIAV), exhibit high genetic diversity and persistent infection in their hosts [2, 5]. ASR has been extensively applied to human immunodeficiency virus (HIV), but the principles are directly transferable to veterinary lentiviruses [2, 7]. Huebert et al. reconstructed ancestral Vif proteins from lentiviral lineages to study the coevolution of Vif with host APOBEC3 restriction factors [2]. The ancestral Vif sequences were functionally tested and shown to antagonize ancient APOBEC3 proteins, revealing a long-standing evolutionary arms race [2]. In FIV, similar ASR studies could identify ancestral envelope glycoprotein conformations that enabled cross-species transmission between felids [2, 5].

Furthermore, ASR has been used to design vaccine immunogens based on ancestral viral sequences [5, 7]. Mesa et al. reconstructed ancestral HIV envelope proteins from an elite neutralizer and demonstrated that these ancestral immunogens elicited broadly neutralizing antibodies in animal models [5]. For veterinary vaccines, ancestral immunogens may offer broader protection against diverse circulating strains of FIV or EIAV [5, 7].

Reconstruction of Replicating Viral Ancestors from Defective Proviruses

In retroviral infections, the majority of proviral DNA is defective due to deletions or hypermutation [6]. Patro et al. developed a method combining multiple-displacement amplification (MDA) with single-genome sequencing to obtain full-length proviral sequences and their integration sites [6]. By comparing two defective proviruses with nonoverlapping deletions from the same clonally expanded cell, they inferred the sequence of the intact parental virus [6]. This "viral reconstruction" approach is a form of ASR that operates at the intrapatient level and can be applied to veterinary retroviruses such as FIV and bovine leukemia virus (BLV) [6]. It allows the characterization of replication-competent ancestors that may have seeded the latent reservoir [6].

Limitations and Challenges

ASR is subject to several limitations that must be considered when interpreting results [1, 3]. First, the accuracy of reconstruction decreases with increasing evolutionary distance and with high substitution rates, which are common in RNA viruses [1, 4]. Second, model misspecification, such as ignoring recombination or selection, can bias ancestral state estimates [2, 6]. Third, ASR assumes that the phylogenetic tree is correct, but tree uncertainty is often not fully propagated into the ancestral reconstruction [3, 7]. Fourth, experimental validation is resource-intensive and may fail if the reconstructed protein is nonfunctional due to epistatic interactions that are not captured by the model [4, 5].

In veterinary contexts, limited sequence data from animal hosts can reduce sampling density and phylogenetic resolution [4]. For emerging viruses, the lack of closely related sequences from the ancestral host may lead to unreliable reconstructions [3]. Despite these challenges, ASR remains a valuable tool when combined with robust phylogenetic methods and functional testing [1, 6].

Future Directions

Advances in computational modeling, including the integration of structural information and deep learning, are expected to improve ASR accuracy [3]. For example, incorporating protein three-dimensional structures into substitution models can account for site-specific constraints on amino acid replacement [2, 4]. Additionally, the use of Bayesian phylogenetic methods that jointly infer tree topology and ancestral states can better accommodate uncertainty [1, 7].

In veterinary virology, ASR will play an increasingly important role in pandemic preparedness by identifying molecular signatures of zoonotic potential in animal viruses [3, 4]. The technique can also guide the design of broadly protective vaccines for livestock and companion animals [5, 7]. As sequencing technologies become more affordable and widespread, the application of ASR to understudied viral families in wildlife and domestic animals will expand [2, 6].

Frequently Asked Questions

What is ancestral sequence reconstruction?

Ancestral sequence reconstruction is a computational method that infers the genetic sequences of extinct or ancestral viruses from contemporary sequence data using phylogenetic trees and models of molecular evolution [1, 2].

How does ASR differ from standard phylogenetic analysis?

Standard phylogenetics focuses on estimating relationships among extant taxa, while ASR explicitly estimates the sequences at internal nodes of the tree, allowing experimental characterization of ancestral proteins [3, 4].

What substitution models are used in ASR?

Common models include the GTR model for nucleotides and empirical amino acid matrices such as JTT, WAG, or LG, often with gamma-distributed rate heterogeneity [2, 5]. Codon-based models are also used for viral sequences under strong selection [3, 4].

Can ASR be applied to any virus?

ASR is applicable to any virus for which a sufficient number of aligned sequences and a reliable phylogeny can be obtained [1, 6]. It is most effective for viruses with moderate substitution rates and clear phylogenetic structure [3, 4].

How is ASR validated experimentally?

Ancestral genes are synthesized, expressed as recombinant proteins or pseudoviruses, and tested in functional assays such as receptor binding, cell entry, and neutralization [4, 5, 7].

What are the main applications of ASR in veterinary virology?

Applications include identifying host adaptation mutations (e.g., influenza receptor binding), reconstructing ancestral glycoproteins for vaccine design, studying host-virus coevolution (e.g., lentiviral Vif-APOBEC3), and inferring replication-competent ancestors from defective proviruses [2, 4, 6].

What are the limitations of ASR?

Limitations include decreased accuracy with high substitution rates, model misspecification, tree uncertainty, and the need for experimental validation [1, 3, 6].

References

[1] Jones BR, Brumme ZL, Hunter E, et al. Getting to the root of HIV transmitted founder virus sequences. Virus Evol. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42110881/

[2] Huebert DNG, Ghorbani A, Lam SYB, et al. Coevolution of Lentiviral Vif with Host A3F and A3G: Insights from Computational Modelling and Ancestral Sequence Reconstruction. Viruses. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40143321/

[3] Brintnell E, Gupta M, Anderson DW. Phylogenetic and Ancestral Sequence Reconstruction of SARS-CoV-2 Reveals Latent Capacity to Bind Human ACE2 Receptor. J Mol Evol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34739551/

[4] Su W, Harfoot R, Su YCF, et al. Ancestral sequence reconstruction pinpoints adaptations that enable avian influenza virus transmission in pigs. Nat Microbiol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34702977/

[5] Mesa KA, Yu B, Wrin T, et al. Ancestral sequences from an elite neutralizer proximal to the development of neutralization resistance as a potential source of HIV vaccine immunogens. PLoS One. 2019. URL: https://pubmed.ncbi.nlm.nih.gov/30969970/

[6] Patro S, Brandt L, Bale MJ, et al. Combined HIV-1 sequence and integration site analysis informs viral dynamics and allows reconstruction of replicating viral ancestors. Proc Natl Acad Sci U S A. 2019. URL: https://www.semanticscholar.org/paper/da02f77b82383d15e0b737444b052ba572fe11a6

[7] Doria-Rose NA, Learn GH, Rodrigo AG, et al. Human immunodeficiency virus type 1 subtype B ancestral envelope protein is functional and elicits neutralizing antibodies in rabbits similar to those elicited by a circulating subtype B envelope. J Virol. 2005. URL: https://pubmed.ncbi.nlm.nih.gov/16103173/ *** Disclaimer: This article is for educational and informational purposes only. It is not intended to substitute for professional veterinary advice, diagnosis, treatment, or regulatory guidance. Always consult a licensed veterinarian or qualified specialist regarding animal health, disease diagnosis, and therapeutic decisions.