Current HIV Research, 2010, 8, 613-621

613

Computational Design of Centralized HIV-1 Genes Miguel Arenas§ and David Posada* Department of Biochemistry, Genetics and Immunology, University of Vigo, E-36310, Vigo, Spain Abstract: The extreme genetic diversity of the HIV-1 remains as a daunting challenge for the development of an effective AIDS vaccine. One strategy for creating a single vaccine that protects against the HIV-1 expanding population is to reconstruct centralized immunogenic sequences that minimize the genetic distance to circulating strains that the vaccine is targeting. Such centralized genes can be estimated with inferred consensus, ancestral and center-of-tree sequences. Although the increased breadth of antibody and T-cell responses induced by the centralized vaccines to date are encouraging, they are modest and may only be partly effective in combating HIV-1. One of the reasons of this limited success might be that several features of HIV-1 molecular evolution have not been yet taken into account in the design of these centralized vaccines, the most important likely being its high recombination rate and complex nucleotide substitution process. Here we describe evolutionary methodologies for the inference of centralized HIV-1 genes, with particular focus on the sources of error introduced by recombination and the model of evolution, in order to foster the development of more effective immunogens before synthesis and assessment in the lab, and final testing in AIDS vaccine trials.

Keywords: Centralized gene, ancestral sequence, consensus, COT, recombination, molecular evolution. INTRODUCTION According to UNAIDS, the Joint United Nations Programme on HIV/AIDS World Health Organization, an estimated 33.4 million people [31.1-35.8 million] worldwide were living with HIV-1 in 2008. An estimated 2.7 million [2.4-3.0 million] became newly infected with HIV-1 and an estimated 2.0 million [1.7-2.4 million] lost their lives to AIDS [1]. The global prevalence of AIDS has stabilized in the last years but important regional differences should be solved [2]. Today, many aspects of HIV-1 pathogenesis remain unclear, and the development of effective anti-HIV-1 drugs, and especially vaccines, has proven very difficult. Progress towards an effective AIDS vaccine, almost thirty years after the identification of HIV-1, remains frustratingly slow [3-6] despite the fact that the development of a safe and preventive HIV-1 vaccine is a global priority [7, 8] especially for those regions of the world with insufficient financial resources to buy antiviral drugs. However, some progresses have been made: a high number of distinct AIDS candidate vaccines are now in various phases of development, and large-scale clinical trials are underway [9]. Among the many considerations that need to be taken into account to develop a good AIDS vaccine [10, 11], one of the most important challenges is the high diversity and rate of evolution of HIV-1 [12-14]. HIV-1 is the fastest evolving organism ever known, due to high mutation and recombination rates coupled with rapid population dynamics [15-17]. Because of this, and since its introduction into the human population approximately 70 years ago, HIV-1 has

*Address correspondence to this author at the Department of Biochemistry, Genetics and Immunology, University of Vigo, E-36310, Vigo, Spain; Tel: 0034 986812038; Fax: 0034 986813828; E-mail: [email protected] §

Current Address: Computational and Molecular Population Genetics Lab (CMPG), Institute of Ecology and Evolution, University of Berne, Baltzerstrasse 6, 3012, Berne, Switzerland. 1570-162X/10 $55.00+.00

developed tremendous genetic diversity [15], resulting in the evolution of multiple distinct clades of viruses that can vary by up to 35% in the envelope protein [e.g., 18]. Several strategies have been developed in order to construct effective AIDS vaccines, using epitopes from circulating-strains, multi-clade immunogens or centralized genes. In this work we review the computational methodology used for the latter. More realistic centralized genes are expected to elicit larger immune responses and therefore are hoped to be more protective. AIDS VACCINES STRATEGIES

BASED

ON

EVOLUTIONARY

Additionally to typical strategies often used for the design of HIV-1 vaccines such as attenuated or inactivated viruses, subunits (recombinant monomeric envelope proteins), live vector-based and DNA plasmids –see an overview of these approaches in Letvin [19]–, evolutionary strategies have been developed where population immunogens are used to compensate for the large diversity of HIV-1, with multiple clades and recombinant forms worldwide. Within these, we can distinguish three main approaches, namely circulating strain, multi-clade and centralized vaccines. The latter is the focus of this review and it will be described in detail in the next section. Circulating-Strain Vaccines A primary concern in designing protective AIDS vaccines is the optimal choice of strains likely to provide protection against the expanding population of HIV-1 variants [18, 20-22]. Here, evolutionary strategies need to be taken into account [23-25]. According to this, several methods for choosing a vaccine candidate on the basis of genetic or protein sequence data have been put forth during recent years. First, the approach followed in current clinical trials is to choose one or a small number of laboratory-grown © 2010 Bentham Science Publishers Ltd.

614 Current HIV Research, 2010, Vol. 8, No. 8

or primary viral isolates, that approximate a “circulating” strain or to simply match the HIV-1 subtype in the targeted population [11, 26-31]. An advantage of this approach is that it should product antigens likely to adopt “native” conformations because it typically employs viral genes derived from a viable virus [32]. However, because of the tremendous levels of HIV-1 diversity, any circulating strain will be genetically, and therefore, antigenically, very different to other epidemiologically unlinked strains likely to infect the host population. In this case, vaccines based on specific viral isolates will not be effective against a broad range of circulating viruses unless key epitopes are conserved. The results of the first phase III AIDS vaccine trial suggested that monomeric envelope proteins that are derived from such isolates were insufficient to provide protective immunity [33]. However, a recent study showed that highly conserved regions in Gag, Env and Nef that are very representative of circulating HIV-1 strains can produce relevant cytotoxic T lymphocyte (CTL) responses [34].

Arenas and Posada

content of a given sample of nucleotide or amino acid sequences. The idea is that these centralized genes can encompass the immunogenetic properties of the diverse circulating strains in the target population [e.g., 32, 42, 43-47]. Centralized genes are currently built with consensus (CON), ancestral (ANC) and center-of-tree (COT) sequences [48] (Fig. 1). The following is a description of the most relevant CON, ANC and COT-based vaccine studies. An overview of the progress in this regard is shown in Table 1. Consensus Genes The most common approach for the construction of centralized genes has been the use of consensus sequences (CON), composed of the most abundant nucleotides or amino acids at each position in the alignment according to a given threshold or frequency. CON sequences are hoped to cover more “conserved” epitopes, and they can be estimated from circulating strains or from strains available in the HIV1 database [12, 21, 49, 50].

Multi-Clade Vaccines Ancestral Genes To enhance the breadth of the elicited immune response, another approach is to include components from as many diverse HIV-1 isolates as possible in the vaccine, with the intention of inducing multiple responses against divergent viral proteins [27, 35]. Indeed, some results support the idea that multi-clade vaccines can elicit robust cellular and humoral immune responses to all vaccine-encoded antigens, with no evidence of antigenic interference [36-41]. CENTRALIZED VACCINES Centralized vaccines are based on the use of “centralized” genes or sequences, which somehow encompass the genetic

While CON sequences are genetically closer to circulating strains than any given natural virus isolate, they can be heavily biased by the particular sample of strains used to obtain the consensus (see below). Also, they can contain polymorphisms in combinations not found in any natural or viable virus, potentially resulting in inappropriate structural conformations. In addition, CON sequences do not consider the evolutionary history of the sample, essential to study biological interactions like those affecting covarying sites [51]. Because of this, others have proposed the use of ancestral sequences (ANC) as vaccine candidates [e.g., 21, 48, 52-54]. The ANC sequence or most recent common

Fig. (1). Consensus (CON), Ancestral (ANC) and Center-Of-Tree (COT) amino acid sequences. CON can be calculated considering the most frequent states at each alignment column, while ANC and COT sequences are inferred using phylogenetic approaches.

Design of Centralized Genes

ancestor (MRCA) of a given sample can be reconstructed from a phylogenetic tree describing the historical relationships of the sampled sequences. In general, the distance between the sequences of the sample and the inferred CON and COT sequences tends to be similar, while the distance to the ANC sequence is usually larger [54-56]. However, the fact that the current HIV-1 diversity has arisen as a result of a series of founder events and progressive divergence implies that a common ancestral sequence may be closer to distant contemporaneous sequences than any of these contemporaneous sequences are to each other [57]. In theory, the amino acid sequence of the MRCA could elicit immune responses that recognize a broad spectrum of viral variants. Center-Of-Tree Genes Finally, another phylogenetic approach to obtain centralized sequences is the center-of-tree method, in theory less sensitive to phylogenetic outliers [53]. COT sequences correspond to a point in the unrooted phylogeny of the target sequences where the average evolutionary distance to each tip on the phylogeny is minimized. Because the COT is a point on the phylogeny, the COT sequences should have similar advantages than ANC sequences. However, depending on the shape of the phylogenetic tree assumed the COT and ANC sequence can be more or less different. In general, the more balanced is the tree topology, the more similar they will be. Because of the way they are constructed, COT sequences will be closest than ANC sequences to rapidly evolving lineages, while they will tend to be equidistant to slow lineages [61]. In order to improve the potential epitope coverage, Nickle et al. [58] suggested the addition of highly frequent variable, but natively preserved sites to the COT sequences. Although the resulting COT+ sequences display a high number of known CTL epitopes, even significantly more than any combination of circulating strains, Fisher et al. [59, 60] argued that the COT+ method, which provides incomplete sequences, could result in a reduction of the population coverage and in the generation of rare and/or unnatural antigens. They suggested instead the use of optimized full-length mosaic proteins [59, 60]. These are composed of fragments of viral sequences and have been optimized using in silico recombination to obtain the maximum coverage of potential T-cell epitopes for a viral population. Nickle et al. [61] compared these methods, arguing that a combinatorial optimization of COT can attain higher coverage than the mosaics, which can also result in unnatural peptides. Sampling Biases Sampling biases is a fundamental aspect that must be considered in the reconstruction of centralized sequences if we want them to be representative of the target viral population. Ideally, the sample should be representative of the phylogenetic distribution of the circulating strains and of their relative frequencies. CON and COT sequences are especially dependent on the sampling process, while ANC sequences are unlikely to change much as new sequences are added to the sample [62]. If a sample contains much more sequences from clade A than from clade B, the resulting

Current HIV Research, 2010, Vol. 8, No. 8

615

CON sequence will be only representative of the former clade and the COT sequence would be much closer to clade A than to clade B. In general, the average distance between the sampled sequences and the CON/COT sequences would likely be lower than to the ANC sequence, but the variance should be higher since at least some of the distances between the clade B and the CON/COT sequence would be large. Efficacy of Centralized Genes The CON approach was initially tested using a group M consensus immunogen (CON6) and shown to elicit broad Tcell responses and weak neutralizing antibody in small animal models [42]. Later on, the CON6-derived vaccine has been shown to induce a greater number of T-cell epitope responses than any single wild-type subtype A, B, and C env immunogen and similar T-cell responses to a polyvalent vaccine [63]. In addition, Malm et al. [64] have showed cross-clade protection by HIV-1 immunogens expressing consensus sequences of multiple genes and epitopes from subtypes A, B, C, and FGH. Moreover, immunogenic Tat, Rev and Nef DNA vaccines derived from subtype C have shown very effective expression [65]. Recent studies confirm that CON sequences are highly representative of Gag, Env and Nef clades [34], and that CON subtype B and C immunogens can be significantly more potent than wild-type vaccines at eliciting neutralizing antibodies and cellular immune responses [44, 46, 66]. Furthermore, Doria-Rose [32] designed a prototype HIV-1 envelope vaccine using ANC sequences. They constructed a full-length ancestral subtype B HIV-1 env gene (An1-EnvB) that produced a glycoprotein that bound and fused with cells expressing the HIV-1 coreceptor CCR5. This artificial gene induced immunoglobulin G antibodies capable of weakly neutralizing heterologous primary HIV-1 strains in rabbits. Later, ancestral and consensus sequences have been predicted to build a specific vaccine for HIV-1 subtype C (called AncC and ConC, respectively) [43]. Both AncC and ConC env genes expressed functional envelope glycoproteins that were immunogenic in laboratory animals and elicited humoral and cellular immune responses of comparable breadth and magnitude. Moreover, Bansal et al. [67] demonstrated a broad cross-reactivity of nearly 70% among all the seven Gag immunogens evaluated with CON and ANC sequences. Additionally, Rolland et al. [55] have shown immunogenic COT sequences that elicited antigen-specific T-cell responses in mice. On the other hand, Frahm et al. [54] were able to detect high cytotoxic T cellular response rates in ELISpot assays using clade B and C, and group M centralized peptides. Interestingly, in this study ANC, COT and CON sequences were similarly powerful, but the combination of these approaches detected significantly broader responses. Moreover, the magnitude of these responses was correlated with the genetic distance between the sampled sequences and the tested population. All together, these results on animal models suggest the utility of centralized gene products (CON, ANC, or COT) as components of an AIDS vaccine. Promising results with centralized proteins have been obtained in their use for the design of peptides as reagents for ELISpot assays [54]. The three methodologies (CON, ANC and COT) produced

616 Current HIV Research, 2010, Vol. 8, No. 8

Table 1.

Arenas and Posada

Overview of Recent Research on HIV-1 Centralized Genes. For Each Study, we Show the Strategy Used (Ancestral “ANC”, Consensus “CON” and/or Center-of-Tree “COT”), HIV-1 Group or Subtype, a Brief Summary of Results, and the Corresponding Publication

Type of Centralized Sequence

HIV-1 Group/Subtype

Summary of Results

Refs.

CON

Group M

Immunogens from the single centralized gene generated cellular immune responses with significantly increased breadth compared with immunogens from wild-type virus gene.

[46]

CON

Group M

Neutralizing antibodies elicited by centralized Env were of similar or greater breadth than those elicited by wild-type Envs.

[45]

CON

B

Consensus Env immunogens appeared to be at least as good as, and in some cases better than, wild-type env immunogens for inducing a neutralizing antibody response.

[44]

CON

Subtypes A1, A2, B, C and D

Vaccine delivery of a 16-valent mixture of regions can focus the CTL response against conserved epitopes that are broadly representative of circulating HIV-1 strains.

[34]

CON

All

Results confirmed the utility of bioinformatics tools to design novel vaccines based on immunogenic consensus sequence. Results were confirmed by ELISpot assays.

[108]

CON

All

Results showed the utility of bioinformatics tools to construct novel immunogenic consensus sequence. Highly immunogenicity was confirmed by ELISpot assays.

[109]

CON

Subtype B

Moderate subtype B-specific antibody response. CON Env was up to four times more potent at driving cellular immune responses and showed increased breadth and magnitude of crossreactive cellular responses.

[110]

CON

All

It shows the utility of bioinformatics tools to design novel vaccines containing immunogenic consensus sequences.

[111]

CON

Group M

CON env genes expressed envelope glycoproteins that retain the structural, functional, and immunogenic properties of wild-type HIV-1 envelopes.

[42]

CON

Subtypes A, B, C, and FGH

Weak humoral immunity was observed but the cross-clade protection observed demonstrated that multigene/multiepitope HIV DNA immunogens were likely to be potent immunogens against HIV-1 infection.

[64]

CON

Group M

The CON vaccine was immunogenic and induced a greater number of T-cell epitope responses, and with higher magnitude, than any single wild-type Env immunogen and similar T-cell responses to a polyvalent vaccine.

[63]

CON

Subtype C

The estimated HIV-1 C CON proteins resulted widely representative for different geographic samples and genomic regions.

[50]

CON

Subtypes B and C

The CON B and C Env vaccines elicited cross-reactive cellular immune responses to epitopes in other clades.

[66]

CON

Subtype A

CON sequences highly covered distinct viral compartments and were largely representative of the major circulating viral strain.

[49]

CON

CRF01_AE

Gag p17 and Env-V3 CON sequences of samples of 9-6 years postinfection in Thailand patients showed high epitope conservation.

[112]

CON, ANC

Subtype C

Both ANC and CON Env genes expressed functional envelope glycoproteins that were immunogenic in laboratory animals and elicited humoral and cellular immune responses of similar breadth and magnitude.

[43]

CON, ANC, COT

Subtypes B, C and Group M

All tested centralized antigens were equally powerful detecting T cell responses in infected individuals. Combinations of the centralized sequences significantly increased detection levels by up to threefold.

[54]

CON, ANC, COT

All (intra-individual)

ANC sequences diverged and were not significantly better than extant sequences to minimize genetic distances at later stages of infection and disease.

[56]

ANC

Subtype B

ANC Env gene produced a glycoprotein that bound and fused with cells expressing the HIV-1 coreceptor CCR5. This Env sequence was also functional in a pseudotyped virus assay. Artificial gene induced immunoglobulin G antibodies capable of weakly neutralizing heterologous primary HIV-1 strains.

[32]

COT

Subtype B

COT+ antigens covered the variation found in many independent HIV-1 isolates into lengths suitable for vaccine immunogens.

[58]

COT

Subtypes B and C

Protein mosaics generated more representative centralized sequences than those sequences obtained with the COT+ approach. Results indicated high potential epitope coverage.

[60]

COT

Subtype B

It is claimed that the COT+ approach is fast and requires low computational cost. Indeed, COT+ sequences could provide high immunization with full-length viral protein immunogens.

[61]

COT

Subtype B

COT proteins were immunogenic, eliciting antigen-specific cytotoxic T-lymphocyte responses in mice.

[55]

Design of Centralized Genes

Current HIV Research, 2010, Vol. 8, No. 8

617

similar results in the generation of peptides that induce Tcell responses. However, although the increased breadth of T-cell and antibody responses induced by the centralized vaccines to date are encouraging, they are modest and may only be partly effective in combating HIV-1 sequence diversity. Studies of additional centralized immunogens, including consensus and ancestral genes from the same subtype, are necessary to examine the influence of the reconstruction process on their immunogenic properties [43]. Finally, the success of these strategies will be always related to the diversity of viral circulating forms in the population that we aim to protect [see 68], including recombinants, which will be key for the success of the vaccine [69].

phylogenetic estimation [76] and inferences derived [77]. Recombination often results in distinct phylogenetic histories for different partitions of an alignment which can have distinct evolutionary relationships. Therefore, traditional phylogenetic methods, which assume a single binary tree, are often incorrect. While the inference of CON sequences itself is not biased by recombination, the estimation of COT and ANC sequences can be affected. Very recently, it has been shown [78] that recombination can severely mislead the estimation of ANC sequences and the prediction of ancestral epitopes and N-glycosylation sites based on those sequences. Consequently, centralized vaccines based on these kinds of approaches might be less effective than expected simply because the sequences assayed are not the ones intended.

COMPUTATIONAL INFERENCE OF CENTRALIZED GENES

We have analyzed some of HIV-1 group M and subtype B Env data sets with moderate recombination rates. ANC sequences inferred ignoring/considering recombination differed by 3.56-4.69% (91-118 nucleotides) and 4.82-9.24% (39-70 amino acids). In general, when recombination was ignored, the number of epitopes and N-glycosylation sites identified in the inferred ancestral sequences was smaller than when recombination was considered and also, importantly, these sites were not exactly the same. Fig. (2) shows the differences between the inferred ANC and COT sequences when recombination is considered and ignored for a HIV-1 subtype B Pol alignment. Interestingly, in this case COT sequences resulted less sensitive to the recombination bias than ANC sequences, probably because recombination makes it impossible to reconstruct the ancestral states that changed between the grand most recent common ancestor of the sample (GMRCA) and the younger MRCA fragments [78].

Nucleotide and amino acid sequences for centralized genes (CON, ANC, or COT) are predicted in silico, which can later be synthesized on the laboratory. Despite all the attempts to explore the applications of centralized vaccines, only a few studies have focused on the implications of the different computational strategies that actually define the sequences of these centralized genes. Doria-Rose et al. [32] used computer simulations to show that the ancestral state reconstruction can be >95% accurate, and 99.8% accurate when estimating the known inoculum used in an experimental HIV-1 infection study in Rhesus macaques (however, they did not consider recombination; see below). Furthermore, the deduced ancestor gene differed from the set of sequences used to derive the ancestor by an average of 12.3%, while these latter sequences were an average of 17.3% different from each other. In addition, Kesturu et al. [56] derived and analyzed CON, ANC, and COT sequences to represent intra-individual HIV-1 env variants encoding a range of diversities and phylogenetic structures. They found that during the first 5 years of infection CON, ANC, and COT genes effectively minimized the genetic distance to the extant sequences, while later in infection the ANC approach performed worse. Finally, Ross et al. [70] showed that different phylogenetic assumptions, specially tree rooting, had an impact on the accuracy of ancestral sequence reconstruction from 118 HIV-1 env gene sequences. They also showed that CON, ANC, or COT approaches might have implications on the predicted 3D structural properties, the number of epitopes, MHC binding sites and N-glycosylation sites. RECOMBINATION BIAS While all the above studies are necessary and very useful, so far none of them have explicitly taken into account one of the most predominant features of HIV-1 evolution: recombination [17, 71]. Genetic recombination is an integral part of the HIV-1 life-cycle, occurring when reverse transcriptase switches between alternative genomic templates during replication. The recombination rate of HIV-1 is one of the highest of all organisms, with an estimated three recombination events occurring per genome per replication cycle [72], thereby exceeding several times the mutation rate. The discovery that most infected cells harbor two or more different proviruses [73], and the evidence for dual infection [74, 75], set the stage for recombination to play a central role in the generation of HIV-1 diversity. Unfortunately, recombination can seriously mislead

Fig. (2). ANC and COT from recombinant sequences. We analyzed an alignment of 26 HIV-1 B pol sequences with 951 bp from the United Kingdom [105]. The recombination rate (
=4Nrl) estimated with the OmegaMap program [106] was 2.1 b. A maximum likelihood (ML) tree was inferred using PhyML [107]. ASR sequences were estimated with HyPhy [87] (ML joint reconstruction) and COT sequences were estimated with COTseq [53, 55]. In order to consider the recombination, we repeated the above procedure, but independently for each recombinant fragment detected by GARD [100], concatenating afterwards the resulting ANC and COT sequences. The recombination bias was measured as the number of nucleotide differences between the inferred sequences considering and ignoring recombination. ANC and COT differences are respectively shown by black and white bars. The bias is shown comparing all sites and only segregating sites.

618 Current HIV Research, 2010, Vol. 8, No. 8

THE INFLUENCE OF THE MODEL OF EVOLUTION On the other hand, the model of evolution used during the phylogenetic analysis might also have an effect on the estimation of ANC and COT sequences. It is clear that models of nucleotide, codon or amino acid substitution play a significant role in phylogenetic estimation, particularly in the context of distance, maximum likelihood and Bayesian estimation [79]. The use of one or other model affects many, if not all, stages of phylogenetic inference, such as estimates of topology, substitution rates, bootstrap values, posterior probabilities, or tests of the molecular clock [e.g., 80, 81]. Ancestral sequence reconstruction (ASR) is also particularly sensitive to model choice [82]. The statistical selection of best-fit models of evolution can reduce the possibility of an incorrect result that best fits the data at hand [83]. Therefore, in order to obtain accurate ANC and COT sequences it should be important to use appropriate models of evolution, especially when we know that different HIV-1 data sets are best fit by distinct models of evolution [84]. METHODS AND SOFTWARE FOR THE ESTIMATION OF CENTRALIZED SEQUENCES The estimation of CON sequences is trivial, and many programs implement it, such as BioEdit (http://www.mbio. ncsu.edu/BioEdit/bioedit.html), JalView [85], Mesquite [86], HyPhy [87] or ConsensusMaker at Los Alamos National Laboratory (http://www.hiv.lanl.gov/content/sequence/CON SENSUS/consensus.html). For the inference of ANC and COT sequences, a phylogenetic tree must be estimated beforehand –although in Bayesian approaches we integrate over many trees to account for phylogenetic uncertainty. As mentioned above, the accuracy of the estimated tree is important, as different trees can result in different predictions (see below). In order to obtain more robust inferences it will be important to use model-based strategies such as Maximum likelihood (ML) and Bayesian approaches [88], and statistically justified models of nucleotide and amino acid evolution [89, 90]. Several methods for the reconstruction of ancestral sequences have been implemented in programs like PAUP* [91], HYPHY [87], PAML [92], MrBayes [93], FASTML [94], GASP [95], ANCESCON [96] and in the Datamonkey (http://www.datamonkey.org/) [97] and Ancestors 1.0 (http://ancestors.bioinfo.uqam.ca/ ancestorWeb/) [98] web servers. Programs like FASTML, GASP and Ancestors are able to reconstruct insertions and deletions in the ANC sequence, which can be considered a necessary step towards reality. On the other hand, COT sequences can be reconstructed with HYPHY or using the DIVEIN server (http://indra.mullins.microbiol.washington.edu/ DIVEIN/cot.html) [99]. The particular aspects related to the reconstruction of ANC sequences (i.e., indel inference) can be also translated to the reconstruction of COT sequences. As mentioned above, recombination poses a problem for the estimation of ANC and COT sequences. These cases require the use of particular methodologies in order to reduce the biases introduced by recombination [78]. For instance, for the inference of ANC and COT sequences, the first step might consist in the detection of the recombinant fragments of the sample, for example with GARD [100], RDP [101], SCUEAL [102] and others (see Posada [71]). Then, a

Arenas and Posada

phylogenetic tree could be estimated for each recombinant fragment, and used for the inference of partial ANC sequences (that is, one for each recombinant fragment), that can be later concatenated into a complete sequence. The Hyphy package and the Datamonkey web server have recently automated this whole procedure from the estimation of ANC sequences in the presence of recombination. However, it is important to note that this concatenated sequence will not represent a real sequence in most instances. This is because in the presence of recombination, different fragments can coalesce at different times, and will have different MRCAs [78]. Nevertheless, the fact that the concatenated ANC maybe did not exist as such, does not preclude its potential use as a centralized sequence, as it contains “real” ancestral fragments. A similar procedure could be used for the COT sequence in the presence of recombination, but as far as we know, this has not been implemented yet. CONCLUSIONS Different strategies exist for the estimation of centralized genes that could be of use for the design of more effective vaccines. Indeed, they all rest on the assumption that the immunogenic epitopes of the virus will be constant to some extent, so the evolutionary arms race between the virus population and the host immune system [103] does not preclude the conservation of some key epitopes. Indeed, the size and composition of the ideal target population will depend upon this co-diversification. While the assessment in the lab of potential centralized immunogens is indeed essential, the design of centralized genes could be further optimized for faster and larger expression in human cells, for example by considering codon usage, effective translation of the transcripts, and protein free energy [e.g., 104]. We also believe that better computational strategies for the design of centralized HIV-1 sequences should be investigated. In particular, HIV-1 recombination needs to be taken into account, while more sophisticated phylogenetic approaches, including the justification of the model of evolution, should result in more accurate inferences. A key issue in the construction of centralized vaccines is the sample that we use to estimate these centralized genes. Indeed, the samples used with such purpose should provide a good representation of the genetic diversity in the target population, and in the case of CON sequences this includes a good estimate of allele frequency and phylogenetic distribution. If the samples are too small, centralized genes can be biased towards particular strains and therefore could fail to provide an effective protection. Moreover, it is very important to choose a representative target population, which could be based on a diverse set of parameters (e.g., regional location of viral isolation, year of isolation, HIV clade or the specific coreceptor usage) [18]. Such a vaccine might be potentially built using the “central” of all known HIV-1 strains, of an HIV-1 sequence subtype, or of viruses circulating in a given geographic region or risk group. Here we have reviewed the current computational methodology for the inference of centralized sequences. We focused on the sources of variation and biases produced from unrealistic assumptions that can bias these inferences, in particular the high recombination and substitution rates

Design of Centralized Genes

observed in HIV-1. More realistic models of HIV-1 evolution (e.g., considering the protein activity) could result in the design of more accurate centralized sequences, which in turn could provide higher epitope coverage and T cellular responses. ACKNOWLEDGEMENTS M.A. was supported by an FPI fellowship BES-20059151 (Spanish Ministry of Education and Science [MEC]). D.P. was partially supported by MEC grant BIO2007-61411. We want to thank the reviewers for their excellent comments.

Current HIV Research, 2010, Vol. 8, No. 8

[25] [26]

[27] [28] [29]

[30]

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9]

[10] [11] [12] [13]

[14] [15]

[16] [17] [18] [19] [20] [21] [22]

[23] [24]

UNAIDS. AIDS epidemic update: Joint United Nations Programme on HIV/AIDS World Health Organization; 2009. Kilmarx PH. Global epidemiology of HIV. Curr Opin HIV AIDS 2009; 4: 240-6. Plotkin SA. Sang Froid in a time of trouble: is a vaccine against HIV possible? J Int AIDS Soc 2009; 12: 2. Wilson NA, Watkins DI. Is an HIV vaccine possible? Braz J Infect Dis 2009; 13: 304-10. Andersson J. HIV after 25 years: how to induce a vaccine? J Intern Med 2008; 263: 215-7. Kawalekar OU, Shedlock DJ, Weiner DB. Current strategies and limitations of HIV vaccines. Curr Opin Investig Drugs 2010; 11: 192-202. Veljkovic V, Veljkovic N, Glisic S, Ho MW. AIDS vaccine: efficacy, safety and ethics. Vaccine 2008; 26: 3072-7. Barouch DH. Challenges in the development of an HIV-1 vaccine. Nature 2008; 455: 613-9. Bansal GP, Malaspina A, Flores J. Future paths for HIV vaccine research: Exploiting results from recent clinical trials and current scientific advances. Curr Opin Mol Ther 2010; 12: 39-46. Harari A, Pantaleo G. Understanding what makes a good versus a bad vaccine. Eur J Immunol 2005; 35: 2528-31. Douek DC, Kwong PD, Nabel GJ. The rational design of an AIDS vaccine. Cell 2006; 124: 677-81. Korber B, Gaschen B, Yusim K, Thakallapally R, Kesmir C, Detours V. Evolutionary and immunological implications of contemporary HIV-1 variation. Br Med Bull 2001; 58: 19-42. Slobod KS, Bonsignori M, Brown SA, Zhan X, Stambas J, Hurwitz JL. HIV vaccines: brief review and discussion of future directions. Expert Rev Vaccines 2005; 4: 305-13. Korber BT, Letvin NL, Haynes BF. T-cell vaccine strategies for human immunodeficiency virus, the virus with a thousand faces. J Virol 2009; 83: 8300-14. Zhang M, Foley B, Schultz AK, et al. The role of recombination in the emergence of a complex and dynamic HIV epidemic. Retrovirology 2010; 7: 25. Coffin JM. HIV population dynamics in vivo: implications for genetic variation, pathogenesis, and therapy. Science 1995; 267: 483-9. Shriner D, Rodrigo AG, Nickle DC, Mullins JI. Pervasive genomic recombination of HIV-1 in vivo. Genetics 2004; 167: 1573-83. McBurney SP, Ross TM. Viral sequence diversity: challenges for AIDS vaccine designs. Expert Rev Vaccines 2008; 7: 1405-17. Letvin NL. Progress in the development of an HIV-1 vaccine. Science 1998; 280: 1875-80. Nabel G, Makgoba W, Esparza J. HIV-1 diversity and vaccine development. Science 2002; 296: 2335. Gaschen B, Taylor J, Yusim K, et al. Diversity considerations in HIV-1 vaccine selection. Science 2002; 296: 2354-60. Kraft Z, Strouss K, Sutton WF, et al. Characterization of neutralizing antibody responses elicited by clade A envelope immunogens derived from early transmitted viruses. J Virol 2008; 82: 5912-21. Mosier DE. HIV-1 envelope evolution and vaccine efficacy. Curr Drug Targets Infect Disord 2005; 5: 171-7. de Oliveira T, Salemi M, Gordon M, et al. Mapping sites of positive selection and amino acid diversification in the HIV

[31]

[32]

[33] [34] [35]

[36] [37]

[38]

[39]

[40]

[41] [42]

[43] [44]

[45]

[46] [47]

[48]

619

genome: an alternative approach to vaccine design? Genetics 2004; 167: 1047-58. Goulder PJ, Watkins DI. HIV and SIV CTL escape: implications for vaccine design. Nat Rev Immunol 2004; 4: 630-40. Esparza J, Osmanov S, Pattou-Markovic C, Toure C, Chang ML, Nixon S. Past, present and future of HIV vaccine trials in developing countries. Vaccine 2002; 20: 1897-8. Francis DP, Gregory T, McElrath MJ, et al. Advancing AIDSVAX to phase 3. Safety, immunogenicity, and plans for phase 3. AIDS Res Hum Retroviruses 1998; 14(Suppl 3): S325-31. Graham BS. Clinical trials of HIV vaccines. Annu Rev Med 2002; 53: 207-21. Pitisuttithum P, Nitayaphan S, Thongcharoen P, et al. Safety and immunogenicity of combinations of recombinant subtype E and B human immunodeficiency virus type 1 envelope glycoprotein 120 vaccines in healthy Thai adults. J Infect Dis 2003; 188: 219-27. Nkolola JP, Essex M. Progress towards an HIV-1 subtype C vaccine. Vaccine 2006; 24: 391-401. Kaur G, Mehra N. Genetic determinants of HIV-1 infection and progression to AIDS: immune response genes. Tissue Antigens 2009; 74: 373-85. 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; 79: 11214-24. Pitisuttithum P. HIV-1 prophylactic vaccine trials in Thailand. Curr HIV Res 2005; 3: 17-30. Yang OO. Candidate vaccine sequences to represent intra- and inter-clade HIV-1 variation. PLoS One 2009; 4: e7388. Heeney JL. Requirement of diverse T-helper responses elicited by HIV vaccines: induction of highly targeted humoral and CTL responses. Expert Rev Vaccines 2004; 3: S53-64. Earl PL, Cotter C, Moss B, et al. Design and evaluation of multigene, multi-clade HIV-1 MVA vaccines. Vaccine 2009; 27: 588595. Seaman MS, Leblanc DF, Grandpre LE, et al. Standardized assessment of NAb responses elicited in rhesus monkeys immunized with single- or multi-clade HIV-1 envelope immunogens. Virology 2007; 367: 175-86. Larke N, Im EJ, Wagner R, et al. Combined single-clade candidate HIV-1 vaccines induce T cell responses limited by multiple forms of in vivo immune interference. Eur J Immunol 2007; 37: 566-77. Seaman MS, Xu L, Beaudry K, et al. Multiclade human immunodeficiency virus type 1 envelope immunogens elicit broad cellular and humoral immunity in rhesus monkeys. J Virol 2005; 79: 2956-63. Kong WP, Huang Y, Yang ZY, Chakrabarti BK, Moodie Z, Nabel GJ. Immunogenicity of multiple gene and clade human immunodeficiency virus type 1 DNA vaccines. J Virol 2003; 77: 12764-72. Chakrabarti BK, Ling X, Yang ZY, et al. Expanded breadth of virus neutralization after immunization with a multiclade envelope HIV vaccine candidate. Vaccine 2005; 23: 3434-45. Gao F, Weaver EA, Lu Z, et al. Antigenicity and immunogenicity of a synthetic human immunodeficiency virus type 1 group M consensus envelope glycoprotein. J Virol 2005; 79: 1154-63. Kothe DL, Li Y, Decker JM, et al. Ancestral and consensus envelope immunogens for HIV-1 subtype C. Virology 2006; 352: 438-49. Kothe DL, Decker JM, Li Y, et al. Antigenicity and immunogenicity of HIV-1 consensus subtype B envelope glycoproteins. Virology 2007; 360: 218-34. Liao HX, Sutherland LL, Xia SM, et al. A group M consensus envelope glycoprotein induces antibodies that neutralize subsets of subtype B and C HIV-1 primary viruses. Virology 2006; 353: 26882. Santra S, Korber BT, Muldoon M, et al. A centralized gene-based HIV-1 vaccine elicits broad cross-clade cellular immune responses in rhesus monkeys. Proc Natl Acad Sci USA 2008; 105: 10489-94. McBurney SP, Ross TM. Developing broadly reactive HIV1/AIDS vaccines: a review of polyvalent and centralized HIV-1 vaccines. Curr Pharm Des 2007; 13: 1957-64. Mullins JI, Nickle DC, Heath L, Rodrigo AG, Learn GH. Immunogen sequence: the fourth tier of AIDS vaccine design. Expert Rev Vaccines 2004; 3: S151-9.

620 Current HIV Research, 2010, Vol. 8, No. 8 [49]

[50]

[51] [52] [53] [54]

[55] [56]

[57] [58] [59]

[60] [61]

[62] [63]

[64]

[65]

[66]

[67] [68]

[69] [70]

[71]

Ellenberger DL, Li B, Lupo LD, et al. Generation of a consensus sequence from prevalent and incident HIV-1 infections in West Africa to guide AIDS vaccine development. Virology 2002; 302: 155-63. Novitsky V, Smith UR, Gilbert P, et al. Human immunodeficiency virus type 1 subtype C molecular phylogeny: consensus sequence for an AIDS vaccine design? J Virol 2002; 76: 5435-51. Pagel M. Inferring the historical patterns of biological evolution. Nature 1999; 401: 877-84. Gao F, Korber BT, Weaver E, Liao HX, Hahn BH, Haynes BF. Centralized immunogens as a vaccine strategy to overcome HIV-1 diversity. Expert Rev Vaccines 2004; 3: S161-8. Nickle DC, Jensen MA, Gottlieb GS, et al. Consensus and ancestral state HIV vaccines. Science 2003; 299: 1515-8; author reply -8. Frahm N, Nickle DC, Linde CH, et al. Increased detection of HIVspecific T cell responses by combination of central sequences with comparable immunogenicity. Aids 2008; 22: 447-56. Rolland M, Jensen MA, Nickle DC, et al. Reconstruction and function of ancestral center-of-tree human immunodeficiency virus type 1 proteins. J Virol 2007; 81: 8507-14. Kesturu GS, Colleton BA, Liu Y, et al. Minimization of genetic distances by the consensus, ancestral, and center-of-tree (COT) sequences for HIV-1 variants within an infected individual and the design of reagents to test immune reactivity. Virology 2006; 348: 437-48. Rambaut A, Posada D, Crandall KA, Holmes EC. The causes and consequences of HIV evolution. Nature Review Genetics 2004; 5: 52-61. Nickle DC, Rolland M, Jensen MA, et al. Coping with viral diversity in HIV vaccine design. PLoS Comput Biol 2007; 3: e75. Fischer W, Perkins S, Theiler J, et al. Polyvalent vaccines for optimal coverage of potential T-cell epitopes in global HIV-1 variants. Nat Med 2007; 13: 100-6. Fischer W, Liao HX, Haynes BF, Letvin NL, Korber B. Coping with viral diversity in HIV vaccine design: a response to Nickle et al. PLoS Comput Biol 2008; 4: e15; author reply e25. Nickle DC, Jojic N, Heckerman D, et al. Comparison of immunogen designs that optimize peptide coverage: reply to Fischer et al. PLoS Comput Biol 2008; 4: e25. Thomson SA, Jaramillo AB, Shoobridge M, et al. Development of a synthetic consensus sequence scrambled antigen HIV-1 vaccine designed for global use. Vaccine 2005; 23: 4647-57. Weaver EA, Lu Z, Camacho ZT, et al. Cross-subtype T-cell immune responses induced by a human immunodeficiency virus type 1 group M consensus env immunogen. J Virol 2006; 80: 674556. Malm M, Rollman E, Ustav M, et al. Cross-clade protection induced by human immunodeficiency virus-1 DNA immunogens expressing consensus sequences of multiple genes and epitopes from subtypes A, B, C, and FGH. Viral Immunol 2005; 18: 678-88. Scriba TJ, zur Megede J, Glashoff RH, Treurnicht FK, Barnett SW, van Rensburg EJ. Functionally-inactive and immunogenic Tat, Rev and Nef DNA vaccines derived from sub-Saharan subtype C human immunodeficiency virus type 1 consensus sequences. Vaccine 2005; 23: 1158-69. McBurney SP, Ross TM. Human immunodeficiency virus-like particles with consensus envelopes elicited broader cell-mediated peripheral and mucosal immune responses than polyvalent and monovalent Env vaccines. Vaccine 2009; 27: 4337-49. Bansal A, Gough E, Ritter D, et al. Group M-based HIV-1 Gag peptides are frequently targeted by T cells in chronically infected US and Zambian patients. Aids 2006; 20: 353-60. Vidal N, Mulanga-Kabeya C, Nzilambi N, Delaporte E, Peeters M. Identification of a complex env subtype E HIV type 1 virus from the democratic republic of congo, recombinant with A, G, H, J, K, and unknown subtypes. AIDS Res Hum Retroviruses 2000; 16: 2059-64. Nájera R, Delgado E, Pérez-Alvarez L, Thomson MM. Genetic recombination and its role in the development of the HIV-1 pandemic. Aids 2002; 16(Suppl 4): S3-16. Ross HA, Nickle DC, Liu Y, et al. Sources of variation in ancestral sequence reconstruction for HIV-1 envelope genes. Evolutionary Bioinformatics Online 2006; 2: 18-41. Posada D. Evaluation of methods for detecting recombination from DNA sequences: empirical data. Mol Biol Evol 2002; 19: 708-17.

Arenas and Posada [72]

[73] [74] [75] [76] [77] [78] [79] [80]

[81] [82] [83]

[84] [85]

[86] [87] [88] [89] [90] [91] [92] [93] [94]

[95] [96] [97] [98]

[99] [100]

[101] [102]

Zhuang J, Jetzt AE, Sun G, et al. Human immunodeficiency virus type 1 recombination: rate, fidelity, and putative hot spots. J Virol 2002; 76: 11273-82. Jung A, Maier R, Vartanian JP, et al. Multiply infected spleen cells in HIV patients. Nature 2002; 418: 144. Jost S, Bernard MC, Kaiser L, et al. A patient with HIV-1 superinfection. N Engl J Med 2002; 347: 731-6. Koelsch KK, Smith DM, Little SJ, et al. Clade B HIV-1 superinfection with wild-type virus after primary infection with drug-resistant clade B virus. Aids 2003; 17: F11-6. Posada D, Crandall KA. The effect of recombination on the accuracy of phylogeny estimation. J Mol Evol 2002; 54: 396-402. Schierup MH, Hein J. Consequences of recombination on traditional phylogenetic analysis. Genetics 2000; 156: 879-91. Arenas M, Posada D. The effect of recombination on the reconstruction of ancestral sequences. Genetics 2010; 184: 1133-9. Sullivan J, Joyce P. Model selection in phylogenetics. Annu Rev Ecol Evol Syst 2005; 36: 445-66. Zhang J. Performance of likelihood ratio tests of evolutionary hypotheses under inadequate substitution models. Mol Biol Evol 1999; 16: 868-75. Lemmon AR, Moriarty EC. The importance of proper model assumption in bayesian phylogenetics. Syst Biol 2004; 53: 265-77. Zhang J, Nei M. Accuracies of ancestral amino acid sequences inferred by the parsimony, likelihood, and distance methods. J Mol Evol 1997; 44(Suppl 1): S139-46. Minin V, Abdo Z, Joyce P, Sullivan J. Performance-based selection of likelihood models for phylogeny estimation. Syst Biol 2003; 52: 674-83. Posada D, Crandall KA. Selecting models of nucleotide substitution: an application to human immunodeficiency virus 1 (HIV-1). Mol Biol Evol 2001; 18: 897-906. Waterhouse AM, Procter JB, Martin DM, Clamp M, Barton GJ. Jalview Version 2--a multiple sequence alignment editor and analysis workbench. Bioinformatics 2009; 25: 1189-91. Maddison WP, Maddison DR. Mesquite: a modular system for evolutionary analysis. 1.05 ed 2004. Kosakovsky Pond SL, Frost SD, Muse SV. HYPHY: Hypothesis testing using phylogenies. Bioinformatics 2005; 21: 676-9. Vandemme A, Salemi M. The Phylogenetic Handbook. Cambridge, UK: Cambridge University Press 2003. Posada D. jModelTest: phylogenetic model averaging. Mol Biol Evol 2008; 25: 1253-6. Abascal F, Zardoya R, Posada D. ProtTest: selection of best-fit models of protein evolution. Bioinformatics 2005; 21: 2104-5. Swofford DL. PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). 4 ed. Sunderland, Massachusetts: Sinauer Associates 2000. Yang Z. PAML 4: phylogenetic analysis by maximum likelihood. Mol Biol Evol 2007; 24: 1586-91. Ronquist F, Huelsenbeck JP. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 2003; 19: 1572-4. Pupko T, Pe'er I, Shamir R, Graur D. A fast algorithm for joint reconstruction of ancestral amino acid sequences. Mol Biol Evol 2000; 17: 890-6. Edwards RJ, Shields DC. GASP: Gapped Ancestral Sequence Prediction for proteins. BMC Bioinformatics 2004; 5: 123. Cai W, Pei J, Grishin NV. Reconstruction of ancestral protein sequences and its applications. BMC Evol Biol 2004; 4: 33. Kosakovsky Pond SL, Frost SD. Datamonkey: rapid detection of selective pressure on individual sites of codon alignments. Bioinformatics 2005; 21: 2531-3. Diallo AB, Makarenkov V, Blanchette M. Ancestors 1.0: a web server for ancestral sequence reconstruction. Bioinformatics 26: 130-1. Deng W, Maust B, Nickle D, et al. DIVEIN: a web server to analyze phylogenies, sequence divergence, diversity, and informative sites. BioTechniques 2010; 48: 405-8. Kosakovsky Pond SL, Posada D, Gravenor MB, Woelk CH, Frost SD. GARD: a genetic algorithm for recombination detection. Bioinformatics 2006; 22: 3096-8. Martin D, Rybicki E. RDP: detection of recombination amongst aligned sequences. Bioinformatics 2000; 16: 562-3. Kosakovsky Pond SL, Posada D, Stawiski E, et al. An evolutionary model-based algorithm for accurate phylogenetic breakpoint

Design of Centralized Genes

[103]

[104]

[105]

[106] [107]

Current HIV Research, 2010, Vol. 8, No. 8

mapping and subtype prediction in HIV-1. PLoS Comput Biol 2009; 5: e1000581. Worobey M, Bjork A, O. WJ. Point, Counterpoint: The Evolution of Pathogenic Viruses and their Human Hosts. Annual Review of Ecology, Evolution, and Systematics 2007; 38: 515-40. Wu X, Jornvall H, Berndt KD, Oppermann U. Codon optimization reveals critical factors for high level expression of two rare codon genes in Escherichia coli: RNA stability and secondary structure but not tRNA abundance. Biochem Biophys Res Commun 2004; 313: 89-96. Hue S, Pillay D, Clewley JP, Pybus OG. Genetic analysis reveals the complex structure of HIV-1 transmission within defined risk groups. Proc Natl Acad Sci USA 2005; 102: 4425-9. Wilson DJ, McVean G. Estimating diversifying selection and functional constraint in the presence of recombination. Genetics 2006; 172: 1411-25. Guindon S, Gascuel O. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 2003; 52: 696-704.

Received: July 14, 2010

[108]

[109]

[110] [111]

[112]

Koita OA, Dabitao D, Mahamadou I, et al. Confirmation of immunogenic consensus sequence HIV-1 T-cell epitopes in Bamako, Mali and Providence, Rhode Island. Hum Vaccin 2006; 2: 119-28. De Groot AS, Bishop EA, Khan B, et al. Engineering immunogenic consensus T helper epitopes for a cross-clade HIV vaccine. Methods 2004; 34: 476-87. Yan J, Yoon H, Kumar S, et al. Enhanced cellular immune responses elicited by an engineered HIV-1 subtype B consensusbased envelope DNA vaccine. Mol Ther 2007; 15: 411-21. De Groot AS, Marcon L, Bishop EA, et al. HIV vaccine development by computer assisted design: the GAIA vaccine. Vaccine 2005; 23: 2136-48. Hamano T, Sawanpanyalert P, Yanai H, et al. Determination of HIV type 1 CRF01_AE gag p17 and env-V3 consensus sequences for HIV/AIDS vaccine design. AIDS Res Hum Retroviruses 2004; 20: 337-40.

Revised: October 4, 2010

PMID: 21054255

621

Accepted: October 10, 2010