Hypervariable antigen genes in malaria have ancient roots

Abstract

Background: The var genes of the human malaria parasite Plasmodium falciparum are highly polymorphic loci coding for the erythrocyte membrane proteins 1 (PfEMP1), which are responsible for the cytoaherence of P. falciparum infected red blood cells to the human vasculature. Cytoadhesion, coupled with differential expression of var genes, contributes to virulence and allows the parasite to establish chronic infections by evading detection from the host's immune system. Although studying genetic diversity is a major focus of recent work on the var genes, little is known about the gene family's origin and evolutionary history.

Results: Using a novel hidden Markov model-based approach and var sequences assembled from additional isolates and species, we are able to reveal elements of both the early evolution of the var genes as well as recent diversifying events. We compare sequences of the var gene DBLα domains from divergent isolates of P. falciparum (3D7 and HB3), and a closely-related species, Plasmodium reichenowi. We find that the gene family is equally large in P. reichenowi and P. falciparum -- with a minimum of 51 var genes in the P. reichenowi genome (compared to 61 in 3D7 and a minimum of 48 in HB3). In addition, we are able to define large, continuous blocks of homologous sequence among P. falciparum and P. reichenowi var gene DBLα domains. These results reveal that the contemporary structure of the var gene family was present before the divergence of P. falciparum and P. reichenowi, estimated to be between 2.5 to 6 million years ago. We also reveal that recombination has played an important and traceable role in both the establishment, and the maintenance, of diversity in the sequences.

Conclusions: Despite the remarkable diversity and rapid evolution found in these loci within and among P. falciparum populations, the basic structure of these domains and the gene family is surprisingly old and stable. Revealing a common structure as well as conserved sequence among two species also has implications for developing new primate-parasite models for studying the pathology and immunology of falciparum malaria, and for studying the population genetics of var genes and associated virulence phenotypes.

Figure 1

A proportional (branch-lengths are not informative) maximum-likelihood phylogeny of amino acid sequences of all DBLα sequences for var genes in P. falciparum isolates 3D7 (PF and MAL, red circles) and HB3 (green circles), and P. reichenowi (Pr, blue circles), including the DBLα region used for reference in this study (DBLa_Su black circles). The tree is a consensus of 1000 bootstrapped replicates, showing only nodes represented in ≥ 70% of the replicate trees.

Figure 2

An example of the output and results of the analytical method. (A) Amino acid alignments are shown for three exemplary target sequences used in the analysis (the first sequence in each alignment is the target) used in the analysis. Each destination sequence is shown for the fragment (for recombinant sequences) or whole sequences (for non-recombinant results, in these examples) that showed the homology match. Each color represents a different amino acid. (B) Mosaic ancestry of the 3D7 DBLα domains. Schematic representations of domains from the 3D7 genome showing the source of the nearest-neighbor of each homology block: 3D7 (red, within genome match, recent homology), HB3 (green, within species match, older homology) and P. reichenowi (blue, across species match, ancient homology). Homology regions are separated by black vertical lines, and adjacent blocks of the same color represent blocks originating from different genes of the same genome. Genes are placed in genomic context showing relative locations to each other and to the telomeres of each chromosome (chromosomes drawn roughly to scale). Possible pseudogenes are marked with *. Only robust, high-scoring alignments are shown. Alignments for 3D7 PF08_0140 and 3D7 PFL0020w are not shown due to poor scores.

Figure 3

Distribution of recombination-block sizes. Both species and all three genomes explored in this study have the same range and mean (11) for distances between recombination breakpoints, which define recombination blocks.

Figure 4

Our analysis shows that recombination is uniform throughout the DBLα domains and does not show a hot- or coldspot structure. This lack of structure is revealed in this histogram showing recombination breakpoint frequency at each amino acid position mapped onto a multiple sequence alignment (white bars). This is overlaid with levels of homology (percent of gaps at each position) at each site (red line). Levels of homology are highly inconsistent over the length of DBLα domains and regions that are highly homologous and align well have no gaps, whereas regions with low homology require many gaps. This figure shows that, correcting for regions where there are large gaps (low homology), there are no large increases or decreases in frequency of recombination breaks typical of hot- or coldspot structure. The multiple sequence alignment is of all DBLα domains from all samples used in this study.