CONSERVED FEATURES OF ANTIGENIC VARIATION

A comparison of the mechanisms of antigenic variation in some of the organisms where it has been investigated supports the view of a polyphyletic origin for this process. At a superficial level there are shared features: reversible, mutually exclusive expression of members of gene family. But investigations of the mechanisms underlying antigenic variation reveal some stark contrasts. A good example of this contrast is the comparison of mechanisms of antigenic variation in the well studied prokaryotic and eukaryotic species, Borrelia hermsii and Trypanosoma brucei, respectively (Donelson, 1995).

If we take a functional rather than mechanistic perspective however, there is considerable similarity between systems used in different organisms. This similarity arises because there is, in my view, a restrictive set of five functional requirements that need to be met for antigenic variation to be of selective advantage to a parasite in evolutionary terms.

Firstly, a microorganism must have the capability to express several different antigens; minimally two. These variable antigens must be immunodominant over non-variable antigens, where the latter are defined as antigens that are identical in all cells in a clonal infection. ‘Immunodominance’ in this context indicates that functional immune responses to lower the numbers of pathogens (by killing them for instance) are directed preferentially to these antigens. Epitopes against which immune responses are generated must differ between variable antigens. The rate of switching between variable antigens per unit of time must be greater than the rate of immune clearance per unit of time.

Second, a microorganism must retain the capability to interact with its environment using non-variable antigens that must be immunologically ‘silent’ relative to variable antigens. To exploit the hosts' resources the parasite will need, for example, a glucose transporter to acquire glucose.

Third, individual cells within an infection must express variable antigens in a mutually exclusive manner. In the simplest scenario, if a cell expresses two variable antigens simultaneously, then the rate of immune clearance will be the same as if it had expressed only one of them. Thus the potential advantage of having the capability to express two variable antigens will have been lost.

Fourth, what applies to individual cells applies equally at the population level within an infection. Sub-populations, defined by their different variable antigens, must be expressed to minimise overlap in the timing of their expression (but see below).

Fifth, the pathogen must be capable of population growth within the mammalian host. For antigenic variation to occur successfully in the absence of growth, there would need to be a remarkable co-operativity between microorganisms to co-ordinate timing of switching and the order in which variable antigens were expressed. I find it difficult to conceive as to how such precise co-ordination could be initiated.

Taken together these five functional requirements give rise to four, ‘cardinal signs’ of antigenic variation – chronicity of infection, high rates of switching, hierarchical expression and reversible expression of variable antigens (Turner, 1999).

There are a number of issues that arise from these functional requirements and cardinal signs that should be highlighted in the context of evolution of antigenic variation. There will be a strong evolutionary drive towards divergence between variable antigens, at least for those epitopes that are immunodominant. This means that there is an expectation for much less sequence conservation between members of a gene family underlying antigenic variation than there would be for other gene families, potentially to the extent that identification of genes in any family using simple BLAST-based algorithms may fail. If there are regions or domains of genes that are conserved between family members this could be because of structural constraints on the proteins, lack of involvement in epitopes recognised by the immune system and/or because the proteins may combine a function for antigenic variation with a second function, such as cellular adhesion in malaria parasites.

There must, inevitably, be trade-offs between the first and second conserved features; variation and immunodominance versus conservation for interaction with the environment. It seems intuitively likely that such trade-offs would be important and interesting and yet they have been hardly explored from an evolutionary perspective. Indeed, I find it difficult to conceive of experimental approaches with the tools currently available in any protozoan parasite.

ANTIGENIC SWITCHING – RATES, HIERARCHIES AND REVERSIBILITY

A more tractable line of enquiry has been the investigation of rates of antigenic switching. It seems self evident that the rate of switching must be higher than the immune response rate (when measured in equivalent units of time) although this statement does raise the issue that the latter does not appear to have been quantified for any host-pathogen system. It could be envisaged that, theoretically, a parasite might either modulate its switching rate in response to cues from the host or switch spontaneously at a high rate. In practice, all pathogens that have been studied in this respect appear to use the second approach, and they do so at remarkably high rates – typically greater than 1 in 1000 cells switches in each generation, two to four orders of magnitude higher than ‘background’ mutation rates (Turner, 1999).

I have previously argued that, for trypanosome infections, the very high rates of switching may have evolved as a bifunctional transmission-enhancing strategy that both evades and depresses immune responses (Turner, 1999). This possibility applies to the other parasites where multiple infections of hosts with isolates of overlapping variable antigen repertoires occur. Holoendemic malaria would be an excellent example of such a scenario. Competition between genotypes has the potential to lead to increased parasitaemia and higher switching rates. This result comes about because expression of a large variety of variable antigens at high levels will increase the potential to avoid any pre-existing immunity to some of those antigens and cross-immunity between variable antigens common to both genotypes. Any impairment of functional immunity is potentially of selective advantage to the parasite if it promotes transmission.

There is a trade-off required between causing immunodepression and evading immunity, and one way in which this might be managed is by hierarchical expression as illustrated in Fig. 2. Switching is non-random and those variable antigens that are switched to most frequently will achieve highest prevalence earliest in an infection. These early variants will engender variant-specific immunity first and be eliminated leading to an inverse correlation of variant-specific switch rates and order of elimination. Because switching is a reversible process, there will continue to be many switch events later in infections back to expressions of variants against which effective immune responses are already in place and thus these switches will never be detected. In other words, there is massive redundancy in the system leading to progressively increasing numbers of suicide switch events and a concomitant reduction in the effective rate of switching which is the overall rate minus the suicide switches. The explanation shown in Fig. 2 would potentially explain mutually exclusive expression in pathogen populations (the fourth functional requirement in the previous section).

Fig. 2. A potential explanation of the relationship of hierarchical expression of variable antigens and the rate of antigenic switching. A. Switching between variable antigens is non-random and those that are switched to most frequently will appear first in an infection and generate variant-specific immune responses first. Thus there is predicted to be a negative correlation of VAT-specific switching rate and order of immune clearance. B. The effect of this is that, whilst the overall potential rate of antigenic variation remains high throughout an infection, in practice the effective rate of switching to expression of variants to which an immune response has not yet been generated will decline progressively.

Very few studies have investigated hierarchical rates of switching, as opposed to hierarchical (non-random) expression of variants in an infection. What evidence there is (Table 2) supports the notion of hierarchical switch rates. The per capita values for switching rates may differ between replicate experiments for each species, but the patterns are consistent.

Fig. 3 illustrates an important distinction in hierarchical expression of variable antigens between T. brucei and P. fragile. In African trypanosomes, reversion of expression is observed in vivo when parasites are transferred between hosts as shown. The linear hierarchy is ‘reset’ for each new infection. [It is also not usually quite as predictable as shown in this particular example (Kosinski, 1980).] Such reversion has been detected in other parasites such as G. lamblia in in vitro culture (Nash, Conrad & Merrit, 1990). In P. fragile infections however, the hierarchy is not reset in each new host. In vivo data for P. chabaudi and P. falciparum also indicate no resetting of the hierarchy (MacLean, Pearson & Phillips, 1982; Hommel et al. 1991). Reversion of expression has been shown by other routes – in in vitro culture under positive selection for P. falciparum (by panning using a variant-specific antibody, Roberts et al. 1992), after mosquito transmission for P. chabaudi (MacLean et al. 1987) and by passaging of P. fragile in splenectomised monkeys (Handunetti, Mendis & David, 1987). There may be a fundamental distinction therefore between switching in parasites such as trypanosomes and G. lamblia on the one hand and Plasmodium spp. on the other. The explanation of hierarchical expression shown in Fig. 2 could apply to the former but not to the latter.

Fig. 3. Hierarchical expression and reversion of variant antigen expression compared in T. brucei and P. fragile. ↓, indicates transfer of parasites between hosts. There is hierarchical expression in both species leading to variant antigens being expressed in sequence in an infection. In T. brucei the hierarchy is ‘reset’ each time parasites are transferred to a new host thus demonstrating reversion of expression, but in P. fragile the hierarchy is not reset. Reversion has been demonstrated by other means (see text). Data derived from Gray (1965) and Handunetti et al. (1987). Not all variant antigens were sampled in the recipient hosts infected with T. brucei.

The important conclusion from this comparison is that the data on reversion of variable antigen expression in Plasmodium needs strengthening. The in vivo evidence for non-reversion in P. falciparum is based on a single experiment (Hommel et al. 1991), as is the in vitro demonstration of reversion (Roberts et al. 1992). The in vivo data are supported by the evidence from two other species. Reversion does not appear to have been investigated in P. vivax (del Portillo et al. 2001), or in the phylogenetically related parasite, B. bovis (Alred, 1998).

PERSPECTIVE

I have focused on (1) the link between antigenic variation and transmission, (2) the high rates of switching and (3) on hierarchical expression to illustrate that we know remarkably little as to how and why these work the way they do despite being, in my view, aspects of the process that are central to our understanding of antigenic variation. Experimental testing of (1), the hypothesis linking antigenic variation and transmission, is sorely needed. Underlying (2) and (3) is the issue of mutually exclusive expression. The explanation I offer above may be a partial answer in terms of the within-host population biology of an infection for some pathogens. For other pathogens it may not, although hopefully the observation may help direct future studies more appropriately. At the level of the single cell, we do not understand how mutually exclusive expression is regulated in any protozoan parasite that I am aware of. This is despite the best efforts of many very talented colleagues. Addressing these issues experimentally is extremely difficult work to undertake but it is essential because antigenic variation is such an important determinant of virulence in several major pathogens. Surely a stronger input from evolutionary biologists could only help?