Screening of Recombinant Filarial Antigens in Animal Models: How Can We Shape Protective Antigens?

1. Background
The development of a subunit vaccine against Onchocerca volvulus is now at a stage where recombinant antigens are being tested in animal models. This is a major bottleneck, as the chimpanzee is the only animal which reliably allows development and reproduction of
O. volvulus. It is clear that chimpanzees cannot be used for screens at an early stage for ethical (and also for financial) reasons. Numerous attempts to establish O. volvulus infections in other mammals did not meet with success due to the host specificity of the filaria. This problem has yet to be resolved and we, therefore, must rely on proxi-models, i. e. models which are as close as possible to the host parasite combination we are studying.
The EMCF has chosen a combination of two animal models which should yield basic information relevant for O.volvulus infection. The screen which I will address in detail is based on the use of the filaria Acanthocheilonema viteae in it's natural host, the jird (Meriones unguiculatus). Jirds can be successfully vaccinated with irradiation attenuated L3 of A. viteae or with culture supernatants of these L3. To screen antigens for their protective potential, groups of jirds are immunized with recombinant antigens of O. volvulus and subsequently challenged with L3 of A. viteae to establish whether the induced immune responses kill the larvae or impair the worms' development. One disadvantage to this approach is that the immunization and challenge antigens are derived from different species. In addition, the lack of defined reagents for the characterization of jird immune responses does not allow a detailed analysis of antibody subclasses or a dissection of T cell responses in jirds. The advantage of this system, however, is the ability to follow the worms' development in the natural host until their maturity; this could allow detection of protective immune responses against early and more mature stages of the parasite.
The question of whether antigens of one filarial species can induce cross protection against another species has been addressed in only a few studies. Earlier work by Storey and colleagues in England showed that immunization with irradiated Brugia malayi L3 induced protection against a challenge with not only B. malayi L3, but also with the L3 of Litomosoides carinii. Therefore, crossprotection between different filarial species does exist, but, we do not yet have data on whether crossprotection between O. volvulus and A. viteae occurs. We do expect, however, that both species share a majority of antigens, since a comparison of six
A. viteae genes, which were cloned and expressed by our group, revealed an approximate 80% shared homology between the amino acids of these particular proteins and those of the respective O. volvulus proteins. The presence of B cell epitopes common to both species has also been demonstrated by studies with monoclonal antibodies.

2. Establishment of Protocol
Before the testing of antigens had begun, several experiments were designed to establish a standard protocol. Culture supernatants of A. viteae L3 were used as a model substance. Such supernatants are produced by keeping vector-derived A. viteae L3 for 24 h in tissue culture medium at 37oC, mimicking the conditions of the early period in the vertebrate host. During this time the L3 adapt to the host and produce a number of proteins. Immunization with these supernatants induced 50-60% protection in our earlier studies. A first series of experiments addressed the question of how adjuvants might influence protective immune responses. Jirds were immunized with L3 culture supernatants, in combination with various adjuvants, and then challenged with A. viteae L3. The results clearly showed that a combination of culture supernatants with 1 of 3 adjuvants (QuilS21 [a plant saponin], STP [a synthetic adjuvant containing a block copolymer], and BCG [extracts of Bacillus Calmette Guerin]) had protective effects. Two block copolymers did not induce significant protection while Ribi, a bacterial cell wall component, had a suppressive effect leading to an increase in worm burden. Interestingly, Ribi induced the highest titers of antifilarial IgG and IgM antibodies, showing that antibody production is not necessarily correlated with protection. Thus, the choice of a suitable adjuvant is of prime importance for obtaining protective responses; subsequent immunization work utilized STP as the adjuvant.
Culture supernatants were surprisingly effective in inducing resistance, as a single immunization induced the same degree of protection as two or three injections. In consensus with David Abraham, a decision was made to use two subcutaneous injections of antigen in an interval of 4 weeks, followed by a challenge infection with 70 or 80 A. viteae L3 two weeks later. Animals were observed up to 12 weeks post infection. In addition, blood was obtained on four occasions to observe the antibody responses and the microfilarial densities. Experiments were terminated 12 weeks p. i. when the animals were carefully dissected and the filarial worms present in the tissues were counted and measured.

3. Antigen Screening
After the screening system was established, testing of candidate antigens began. These proteins had been selected according to very stringent criteria, ensuring that antigens with a high chance of being protective were tested. To minimize the chance for misinterpretation of results, individual antigens, rather than antigen cocktails, were used in most cases. Results of the antigen screening thus far are shown in the accompanying table. To this point, no antigen tested in this system has proven to be protective; however, this does not mean that these antigens do not have protective properties. Instead, results suggest that the screening protocol may have some shortcomings.
There are certainly factors which could be changed relatively easily, for example, in the choice of adjuvants. Weil and colleagues showed that a truncated form of recombinant paramyosin from Brugia malayi induced about 50% protection in jirds that were challenged with B. malayi L3 when the immunization was done using Freund's Complete Adjuvant. Similarly, Mark Taylor successfully immunized jirds with two recombinant O. volvulus fusion proteins together with FCA, inducing about 50% resistance against a challenge with A. viteae L3. This adjuvant is very potent but also induces inflammatory responses which lead to pathology; it is, therefore, unacceptable for human use. In fact, German authorities do not allow experimentation with this adjuvant except in very special cases. However, information obtained by groups using FCA could be valuable by shedding light on some of the immune mechanisms leading to protection. A few antigens have been tested in our model with different adjuvants (STP or Freund's Complete Adjuvant) or were tested in different formats (e. g. MBP-fusion protein versus GST-fusion protein), but, thus far, none of these changes appear to have influenced the protective outcome.
Interestingly, the lack of protection demonstrated by the 24 screened antigens does not appear to be due to poor immunogenicity. Serological analysis revealed that most of the antigens induced IgG and IgM responses readily detectable by ELISA. In most cases, a good antibody response against the respective recombinant antigen was already present at the time of challenge infection and increased during infection (see table). This demonstrates that the initially induced antibody response was boosted by the infection (i. e. the worms contained epitopes common to the recombinant antigens) and supports the conclusion that A. viteae and
O. volvulus share epitopes. In some cases, however, the maximal antibody responses were present at the time of challenge and the antibody titers declined rapidly during infection, indicating that no boosting had occurred. This is possibly due to a lack of epitopes common to worm antigens and recombinant polypeptides, or to the fact that the respective worm antigen was not available for stimulation, as is the case with certain "concealed antigens".
The antibody responses of most of the immunized animals appear to share a common trait which we do not yet fully understand. Sera of jirds immunized with recombinant antigens were tested by ELISA for reactivity to the recombinant antigen used in the immunization or against extracts of female A. viteae. In the majority of cases the sera reacted well to the recombinant antigens but showed only weak reactivity to female worm antigen. A possible explanation lies in the fact that E. coli-derived antigens (of which many of the recombinants are) are not glycosylated and probably differ with respect to their post-translational modifications from native worm antigens. Therefore, worm-derived antigens can be expected to carry residues (carbohydrates, lipids, phospolipids etc.) which convey a particular tertiary structure to the protein and perhaps cover certain parts of the protein backbone. Perhaps, antibodies against the "naked" E. coli-expressed antigens do not have sufficient access to certain hidden epitopes of the native worm proteins.

4. Food For Thought
In order to ascertain why the antigens tested thus far do not confer much protection, we need to first compare the results with those from the chamber model and with those from vaccine studies in other helminth systems. There may be several reasons for this lack of protection. First, one obvious possibility is that the scientific community has not yet found the principle antigens which induce protective immunity. Second, the structure and immunological properties of E. coli-expressed antigens could be different from native worm antigens, resulting in immune responses which are inefficient for killing the worms. Third, the means of antigen delivery could be a key element determining whether protective or inefficient immune responses are induced. All of these points should be kept in mind as new research projects are designed.
Comparisons with studies on other anti-helminth vaccines show interesting parallels to the onchocerciasis vaccine project and could tell us what approach we should envisage. In the last few years, the H11 antigen of Haemonchus contortus, an intestinal nematode of sheep, was characterized as a protective antigen. Several studies revealed that the native, glycosylated H11 antigen (a membrane protein from the worm's gut purified from worm material by lectin-affinity) protected sheep against challenge infections. While the vaccinated sheep were not completely protected, they had a drastically reduced egg output, a lowered worm burden, and suffered less from the disease. Thus far, no recombinant protective H11 antigen has been published, and informal information tells us that the recombinant formulations of this antigen were substantially less protective or nonprotective. This suggests that native or near-native forms of filarial antigen might be better than the nonglycosylated E. coli-expressed antigens so far used in our screen.
A second example comes from our own laboratory and has been reported by Volker Mueller. Mueller attempted to vaccinate mice against a challenge infection with eggs of the cestode Echinococcus multilocularis. The recombinant antigen used was glyceraldehyde 3 phospate dehydrogenase (GAPDH), a housekeeping enzyme primarily found in the glycolytic pathway of E. multilocularis. This protein has been described as a vaccine candidate for Schistosoma mansoni, but has never demonstrated protective properties. When E. coli-expressed GAPDH was administered to mice, the immunization induced strong antibody responses, but no protection. However, following immunization of GAPDH via living attenuated Salmonella typhimurium, the animals were significantly protected against a challenge infection, although no antibody responses against GAPDH were detectable. Salmonellae induce predominantly Th1-type T cell responses as has been shown by other groups and it is possible that this property was responsible for turning our otherwise inefficient antigen into a protective protein.
These two examples suggest that we should invest more energy in further evolving the recombinant antigens which we have already characterized. This should of course not discourage the characterization of new promising candidate antigens, but it would probably be unwise to simply look for new antigens, forgetting those which did not induce protection in our initial screens. The tool box of modern immunology contains many new approaches and the Clark Foundation has encouraged the onchocerciasis community to utilize them. Among others, genetic immunization, adjuvant immunotherapy and new antigen delivery systems could be instrumental in helping us to shape candidate antigens.

Summary of recombinant antigens tested in the A. viteae model (Lucius)