Redundancy and recombination in the Echinococcus AgB multigene family: is there any similarity with protozoan contingency genes?

Numerous genetic variants of the Echinococcus antigen B (AgB) are encountered within a single metacestode. This could be a reflection of gene redundancy or the result of a somatic hypermutation process. We evaluate the complexity of the AgB multigene family by characterizing the upstream promoter regions of the 4 already known genes (EgAgB1-EgAgB4) and evaluating their redundancy in the genome of 3 Echinococcus species (E. granulosus, E. ortleppi and E. multilocularis) using PCR-based approaches. We have ascertained that the number of AgB gene copies is quite variable, both within and between species. The most repetitive gene seems to be AgB3, of which there are more than 110 copies in E. ortleppi. For E. granulosus, we have cloned and characterized 10 distinct upstream promoter regions of AgB3 from a single metacestode. Our sequences suggest that AgB1 and AgB3 are involved in gene conversion. These results are discussed in light of the role of gene redundancy and recombination in parasite evasion mechanisms of host immunity, which at present are known for protozoan organisms, but virtually unknown for multicellular parasites.


I N T R O D U C T I O N
Protozoan parasites escape the host's immunological defences by switching their expressed antigens. Antigenic variation mediated by genetic recombination is well documented for protozoan parasites, but analogous mechanisms have never been characterized in multicellular parasites. In African trypanosomes, VSG (Variant Surface Glycoprotein) switching results either from homologous DNA recombination that is targeted to an active expression site, which is mainly gene conversion, or from transcriptional in situ switching between different VSG expression sites (Pays, 2005). The active VSG expression site is considered as a recombination hot-spot that generates genes for new VSGs and for novel VSG-based adaptive molecules. Following the genome sequencing, it was shown that trypanosomes have a high potential to generate novelties in this evolutionary process, since most VSGs are pseudogenes (Berriman et al. 2005).
We have recently uncovered extensive genetic diversity underlying one of the major Echinococcus immunodiagnostic antigens, antigen B (AgB) (Chemale et al. 2001 ;Arend et al. 2004 ;Haag et al. 2004). The polymeric antigen is a complex 160 kDa thermo-stable lipoprotein built by 8 kDa subunits, encoded by at least 4 distinct, already characterized, genes AgB1, AgB2, AgB3 and AgB4 (Shepherd et al. 1991 ;Fernández et al. 1996 ;Chemale et al. 2001 ;Arend et al. 2004). Although several AgB gene variants have been isolated and sequenced, their redundancy and genomic organization is still unknown.
Several studies suggest that AgB has an important role in the parasite-host interaction. For instance, AgB is the first antigen recognized by specific antibodies in sera from mice infected with activated oncospheres (Zhang et al. 2003). Furthermore, AgB seems to be involved in the evasion of the host immune response, as indicated by its ability to inhibit polymorphonuclear cell recruitment (Shepherd et al. 1991 ;Riganò et al. 2001). Studying the immunological profile of human patients with different stages of hydatid disease, it has also been shown that parasite survival is predominantly associated with a Th2 lymphocyte polarization in its host (Riganò et al. 2001(Riganò et al. , 2004. Chemale et al. (2005) have suggested that AgB could also be involved in some kind of detoxifying mechanism, sequestering and buffering xenobiotics, since it does not release lipids in competition assays, and therefore does not seem to direct fatty acid metabolic pathways. Recent studies of AgB expression in E. multilocularis show that AgB3 is constitutively expressed in all parasite stages, but the transcription of AgB1, AgB2 and AgB4 appears to be restricted to the larval stage (Mamuti et al. 2006).
Intriguingly, a single E. granulosus metacestode (cyst), originating from 1 single parasite egg, and thus to be considered as a clone, has been found to express a large number of slightly different variants of AgB2 and AgB4 (Arend et al. 2004). Moreover, cloned PCR fragments of AgB1-AgB4 genes from individual protoscoleces, have shown that inter-protoscoleces genetic variation does not account for the total genetic variability encountered within a single hydatid cyst. A single protoscolex harbours several variants for each gene, although one of them is usually the most frequent (Haag et al. 2004). In the same study, theoretical analyses suggest that the concomitant presence of slightly different AgB variants within a single cyst may be adaptive. As parts of a highly immunogenic protein, AgB subunits clearly evolve by positive selection, but the heterogeneous distribution of the positively selected epitopes among subunits indicate distinct biological roles (Haag et al. 2006).
However, the meaning and origin of AgB diversity is incomplete because all genetic studies have included only the short coding sequences, which are no longer than 450 bp. It remains unclear if AgB diversity is a static reflection of high gene redundancy, in which different gene copies originated by duplication slowly diverge over time, or if genetic variability is generated by some particular mechanism, such as somatic hypermutation, recombination, or both, during the progression of the infection. Here we present an extensive cloning and sequencing study of the upstream flanking regions from the 4 already known E. granulosus genes (EgAgB1-EgAgB4), using long and accurate PCR (LA PCR) (Mukai and Nakagawa, 1996). Additionally, we evaluate the redundancy of each gene in 3 Echinococcus species (E. granulous, E. multilocularis, and E. ortleppi) using a real-time PCR-based approach.

Parasite materials
Total genomic DNA was phenol/chloroformextracted from E. granulosus and E. ortleppi protoscoleces collected from bovine hydatid cysts using standard procedures. The respective species was identified by sequencing a fragment of the mitochondrial cytochrome c oxidase subunit I gene as previously described (Bowles et al. 1992). Additionally, we used some previously isolated DNA from E. multilocularis metacestodes for comparison. One metacestode/cyst was considered as a single individual/isolate.

LA-PCR
Approximately 10 mg of total genomic DNA derived from a single E. granulosus sheep strain metacestode (cyst) was digested with 50 U Spe I (New England) and 2 . 5 mg of the digestion was used for ligation to a Xba I cassette (Takara) at 16 xC for 30 min. One ml of ligation was used in 4 independent PCR reactions, each using a forward primer annealing to the adaptor (C1, Takara) and a reverse primer designed to anneal within the CDS of each of the 4 known E. granulosus AgB genes (Tab. 1 -first step), with the LA Taq polymerase (Takara) for reduced incorporation errors and long templates, following the manufacturer's instructions. Subsequently, 1 ml dilutions (1 : 10, 1 : 100 and 1 : 1000) of the first PCR amplicons were used as templates for a second enrichment-PCR step, using the C2 forward primer annealing to the adaptor and a second reverse primer specific for each AgB gene ( Table 1).
The second PCR products were run in an agarose gel containing ethidium bromide, and stained bands were excised for purification using the Qiagen Gel Extraction Kit (Qiagen). One ml from the agaroseeluted DNA was cloned into the TOPO TA Vector for sequencing (Invitrogen). The recombinant plasmids were purified with columns (Qiagen) and sequenced using M13 (-20) forward and M13 reverse primers (Invitrogen) with the Big Dye Version 2.0 kit on an ABI 377 sequencer (Applied Biosystems).

Real-time PCR
Four primer pairs were designed to specifically amplify the 4 E. granulosus or E. multilocularis AgB Table 1. Echinococcus granulosus AgB specific primers used in the LA-PCR (5k-3k) Gene First step Second step GTGTCCCGACGCATGACTTA GACATATTTCTTCAACACTTCGTGAAC genes. We used a specific forward primer annealing to the 5k region inside the second exon of each gene and species, and a reverse primer annealing to the 3k flanking region of AgB1 and AgB3 or AgB2 and AgB4 in both species (respective primer sequences are shown in Table 2). For comparison, we also amplified a segment of the single copy cytosolic malate dehydrogenase gene (mdh) from each isolate. The amplification was performed on a LightCycler (Roche Diagnostics) using the SYBR Green I as a dsDNA-specific fluorescent dye and a continuous monitoring. PCR was initiated by a ' hot start ' of 15 min at 95 xC followed by 40 cycles of denaturing (94 xC, 15 sec), annealing (50 xC, 30 sec) and extension (72 xC, 30 sec). Fluorescence was measured each cycle at 72 xC. Amplifications of DNA molecules derived from either different AgB genes or the single copy mdh gene were performed in triplicates. Mean values from triplicate determinations for each sample were taken for the calculation of relative AgB copy numbers, using the ratio of AgB to mdh amplification. The relative quantifications were performed taking into account the efficiency of the PCR reactions with templates derived from different primer pairs/species. Efficiency was determined by the slope of the regression curve between the fluorescence crossing points and the values of a dilution series of the respective genomic DNA. A dilution series of the corresponding PCR products was also used to calculate the slope (efficiency), and the results were similar to those obtained with dilutions of genomic DNA.

R E S U L T S
We have used 4 sets of specific primers complementary to the downstream CDS region of each AgB gene in combination with adaptor primers C1 and C2 from the LA PCR kit (Takara) to amplify the unknown upstream AgB flanking regions. After the second enrichment-PCR step, a clear band of about 0 . 9-1 . 0 kb appeared for AgB2 and AgB4, while a complex banding pattern appeared for AgB1 and AgB3 (data not shown), the last target showing several amplicons ranging from 0 . 2 to more than 1 . 0 kb in size. All 47 clones sequenced in both directions showed a clear relation to one of the previously characterized AgB genes in their downstream portion, corresponding to the CDS, as shown by the scores of blast search hits (Table 3). Flanking regions of different clones from the same AgB gene are highly divergent in sequence, except for clones B4-A and B4-B, which differ mainly by the absence of the upstream Spe I restriction site in B4-A, leading to a larger fragment. The most divergent clones could not even be properly aligned, and the similarlity among them was usually decreasing in the upstream direction. Figure 1 shows the alignment of the similar regions of our clones and a Table 2 comparison with the reference AgB1-AgB4 CDS sequences used to design the primers for the LA-PCR strategy. Note that these alignments do not contain the complete clone sequences, but only the regions which could be compared through a multiple alignment.
Searching for consensus promoter motifs within our sequences using a neural network-based tool available at http://www.fruitfly.org/seq_tools/ promoter.html (Roese, 2001), we found that the typical AgB promoter-like element includes a TATAA string 28 bp upstream from a stretch of (GT) n localized just after the putative transcription start point. The number of GT repeats is polymorphic both within and between genes, ranging from 3 in AgB1 to 12-13 in AgB4 (see Fig. 1). Several other putative promoter sequences appear isolated in distinct AgB1 and AgB3 clones, or accompanying a typical element from AgB1 and AgB2 genes (data not shown).
Clone B3-A is a clear recombinant. Its upstream non-coding region is highly similar to an E. multilocularis AgB1 sequence, which is also the best hit for clones B1-A and B1-C, but its downstream CDS region is more similar to an E. multilocularis AgB3 sequence (Table 3). This can be more clearly verified through a dot plot, as presented in Fig. 2.
The downstream region of B3-A is more similar to clone B3-C, whereas its upstream region has its counterpart in clone B1-A. To evaluate if this could be the result of a gene conversion event, we aligned the 8 B1-A and the 2 B3-A sequences using the BioEdit software version 5.0.9 (Hall, 1999) and searched for fragments involved in gene conversion with GENECONV version 1.8.1 (Sawyer, 1989). We found that between sites 88 and 203 the aligned region from the two B3-A clones has a greater similarity with the corresponding region in 3 sequences from the B1-A group, than would be expected by chance (based upon 10 000 permutations ; P<0 . 001 for both global inner and global outer fragments).
The heterogeneity of AgB1 and AgB3 flanking region leads to a question about the degree of redundancy of these genes in the E. granulosus genome. Although our previous work based on genomic Southern blots suggested redundancy (Chemale et al. 2001 ;Haag et al. 2004), the exact number of gene copies was never determined. For this, we decided to use another approach to access the question, using real-time PCR. We compared the number of copies of the 4 AgB genes in relation to a single copy mdh gene in 3 Echinococcus species. Our results show that the genes can be quite variable in numbers, not only between, but also within species (Fig. 3).
AgB2 showed the lowest number of copies in the 3 taxa, whereas AgB3 showed the highest in E. granulosus and E. ortleppi. In this species, the number of AgB3 copies varied extensively, from 7 to 117. The most redundant AgB gene of E. multilocularis is AgB4, but the numbers also vary considerably in this species, from 10 to 69 copies.

D I S C U S S I O N
In the present paper we show that the number of Echinococcus AgB genes is higher than previously thought. Furthermore, our results demonstrate that AgB numbers are quite variable between metacestodes from the same species, indicating that DNA rearrangements might be involved in some sort of gene amplification mechanism. The LA PCR experiments followed by sequencing of the E. granulosus AgB genes showed that about half of the repeated AgB3 sequences do not seem to be functional, since they lack a consensus promoter motif.
As suggested in a previous study (Haag et al. 2004), AgB3-related genes appear to be involved in gene conversion ; some of these apparently non-functional copies could be intermediates or derivatives of recombination processes. Furthermore, we sequenced an AgB3 clone where the upstream region has a higher than expected similarity with an AgB1 clone. This is a more reliable indication of gene conversion than that obtained in our previous study (Haag et al. 2004), where the sequences were derived from the same PCR reaction, and some could have been generated during fragment amplification. In the present case, clones B1-A and B3-A were obtained in independent PCR reactions, suggesting that gene conversion must have taken place in the parasite genomic DNA ; clearly, they are not an experimental artifact.
The selective advantage/functional role of maintaining this kind of diversifying mechanism in the Echinococcus genome is a challenging issue. Since clones B1-A and B1-C contain 2 types of consensus promoter motifs upstream to the coding region, we think that gene conversion could up-or downregulate a particular gene by introducing a new promoter motif within its upstream flanking region. We found that the first promoter motif within clone B1-A (sites 1070-1120 in clone B1-A, corresponding to sites 177-227 in clone B3-A), would generate an alternative protein, unrelated to AgB. Indeed, part of this motif is included within the fragment previously identified to result from gene conversion (sites 88 to 203). Further experimental approaches, such as comparing the functionality and relative strength of both putative promoters, should be performed to understand the relevance of gene conversion in this context. Not only B1-A, B1-C, but also clone B2-A contain 2 consensus promoters in the same orientation as the AgB CDS. Furthermore, most AgB1 and AgB3 putative promoters overlap to consensus promoter motifs with a search score higher than 0 . 9 in the opposite orientation (data not shown). The additional promoter consensus motif of B2-A occurs within the second exon inside the AgB2 CDS. In a previous study (Arend et al. 2004) we cloned an AgB2 PCR fragment lacking the region corresponding to this particular motif ; because the excised region showed conserved splicing signals, we hypothesized that this PCR product was derived from a retrotransposed sequence. EgAgB2derived PCR fragments lacking the intron were found in other cloning experiments (Haag et al. 2004), indicating that this gene is prone to retrotransposition.
In Plasmodium falciparum var genes, it has been shown that the type of promoter associated with a particular gene strongly correlates with its chromosomal localization (in telomeric or internal domains), and that transcriptional control of genes located in different chromosomal regions occurs by distinct mechanisms (Voss et al. 2003). In Trypanosoma brucei, the expression of specific variant surface glycoprotein isoforms also relies upon gene location. Having a large repository of redundant vsg genes, DNA rearrangements are used to replace one isoform by another in the active expression site. However, the ' on ' and ' off ' states of a particular expression site promoter do not seem to be associated with DNA alterations (Borst and Ulbert, 2001). Trypanosoma cruzi, on the other hand, does not show antigen switching, but counts on the high genetic heterogeneity of its populations to evade immunity of a wide variety of hosts (Machado et al. 2006).
There remains much to be learned about how E. granulosus metacestodes confront the host immune system. It is clear that this interaction must be distinct from the well-known immunity escape mechanisms found in protozoans, essentially mediated by antigenic switches. E. granulosus metacestodes do not directly interact with the host cellular immune response, because they are physically protected by the laminated layer. However, antigenic variability might be critical during early intermediate host invasion, when oncospheres must reach the target organ from the digestive tract through the vascular system and establish a successful infection. At later stages, metacestode metabolites released into the environment could orchestrate the periparasitic inflammatory reaction in its own favour, and antigenic variability may represent one way to favour its survival potential. Although there is still no link between AgB diversity, gene conversion processes, and their adaptive significance in terms of immunity evasion, the  analogy with the protozoan parasite escape strategies is tempting.
We thank CNPq, CAPES and FAPERGS for supporting our work and Robin Gasser for the critical reading of the manuscript.

R E F E R E N C E S
Arend, A. C., Zaha, A., Ayala, F. J. and Haag, K. L.