Human schistosomiasis caused by parasitic flatworms of the genus Schistosoma remains an important public health problem in spite of concerted efforts at control. An effective vaccine would be a useful addition to control strategies that currently rely on chemotherapy, but such a product is not imminent. In this review, likely causes for the lack of progress are first considered. These include the strategies used by worms to evade the immune response, concepts that have misdirected the field, an emphasis on internal antigens, and the use of the laboratory mouse for vaccine testing. On a positive note, recent investigations on self-cure by the rhesus macaque offer the most promising context for vaccine development. The identification of proteins at the parasite–host interface, especially those of the esophageal glands involved in blood processing, has provided an entirely new category of vaccine candidates that merit evaluation.
The present study is not intended as a comprehensive review of schistosome vaccines. Rather, it is our appraisal of why, in spite of decades of research and much enthusiasm, there is no product that meets even modest criteria for success. Unless the problems are recognized and addressed, there is little prospect of success, yet there are still underexploited avenues that could lead to the goal of a useful product.
A schistosome vaccine would be a good thing
Recent modeling has shown that a vaccine with ∼80% efficacy, both as a prophylactic to prevent infection and as a therapeutic to eliminate existing worm populations, would be invaluable for the control of schistosomiasis . Each schistosome infection by skin penetration is an independent event with no subsequent multiplication in the human host, cf. viruses, bacteria, and protozoa. The adults are long lived in the portal bloodstream or vesical venous plexus of the bladder, depending on Schistosoma species (mean of 8 years for S. mansoni, ). Praziquantel is a good chemotherapeutic and the mainstay of control programs, but it does not protect against reinfection or kill schistosomula larvae . Moreover, schistosomiasis is cryptic with morbidity roughly proportional to infection intensity . Current diagnostics for intestinal schistosomiasis are relatively insensitive , so fail to detect low-level infections that could still have clinical sequelae. An effective vaccine would plug that treatment gap. The most successful human vaccines have been developed against infections where a single exposure elicits life-long immunity in the survivor (think smallpox, measles, and yellow fever). The complete reverse holds for schistosomiasis where an infection, once acquired, can persist undetected for decades . Reinfection after curative chemotherapy is the norm in children (the most at-risk group)  and for immunity to develop in adults requires prolonged exposure . Indeed, changes in hormonal levels at puberty have been put forward as an alternative explanation for low reinfection rates in humans . It is fair to say that no practicable leads that could inform vaccine development have emerged from numerous studies of human immune responses to the disease, especially if these involve IgE antibodies as the mediating agent .
Schistosomes are smart parasites
A major problem for vaccine development is that schistosomes exist and thrive precisely because, as recently shown, their exposed/secreted proteins have been strongly selected to evade humoral and cellular immune effector responses . They reside in the host bloodstream, surrounded by antibodies and effector leukocytes, and their diet comprises the same components ingested into the blind-ending gut for digestion and absorption. Clearly, they have superb immune evasion strategies (reviewed in ref. ). They possess an external shield in the form of a secreted membranocalyx that protects the plasma membrane of the syncytial surface tegument from immune attack , adsorbing host components such as CD44  and repelling the binding of host leukocytes . The alimentary tract may also be protected from immune attack by the secreted gastrodermal and esophageal hydrolases [13,14], which could rapidly destroy ingested plasma constituents. Schistosomes also deploy an internal shield in the form of a battery of antioxidant enzymes to protect against toxic oxygen and nitrogen products from leukocytes . If those were not sufficient defense, they secrete a variety of novel proteins , akin to bacterial virulence factors, which may be capable of subverting immune attack . Larval parasites migrating from the skin infection site to the portal system are also protected; they acquire the adult tegument surface configuration within 24 h of penetration , possibly in little more than 3 h . The genes encoding gastrodermal enzymes are already active at the day 3 skin stage , even though feeding on blood cells does not begin until after arrival in the portal vein. Finally, their levels of antioxidant proteins increase rapidly after skin penetration .
Misleading concepts have deflected the search for a vaccine
Historically, the concepts of concomitant immunity  and antibody-dependent cellular cytotoxicity (ADCC)  dominated the vaccine field for decades. The concomitant immunity hypothesis held that rhesus macaques with a primary S. mansoni infection were fully protected against a cercarial challenge  but crucially, the persistence of the primary burden was never verified by worm recovery. The situation in rhesus macaques is better described as self-cure. The ADCC paradigm rested on the demonstration that schistosomula, artificially transformed from cercariae in vitro, were highly susceptible to killing by a combination of antibody isotypes and effector leukocytes , a concept that dovetailed neatly with concomitant immunity. Identify the targets of immune attack on the larval surface and you had your vaccine candidates.
Preventing parasite establishment at the point of entry is a very attractive proposition, but there is an almost total lack of histopathological evidence for the occurrence of ADCC against skin-stage schistosomes in vivo in any host (Figure 1). The parasites remain distal, in the epidermis, after skin entry  while, of necessity, exposing internal proteins when shedding cercarial membranes during transformation to schistosomula . The window of susceptibility to ADCC in vitro closes in <24 h as the new nonreactive surface is established . Remaining peripheral in the skin beyond the reach of antibodies and leukocytes may be a positional solution to the problem of transient susceptibility to ADCC, but recent work has shown that the parasite can also actively regulate host dermal hypo-responsiveness by driving IL-10 production [22,23].
Schistosomula in the epidermis are not attacked and killed by leukocytes.
Another limitation of ADCC in the skin against single invading parasites is the need for the immune system to be permanently on red-alert . This is a major conceptual difference from viral or microbial infections as there is simply no time to mount a secondary response during the <24 h window of susceptibility, and the sporadic nature of parasite acquisition means that antibody titers to the cercarial secretions are never high, either in natural infections or after exposure of primate hosts to many thousands of attenuated cercariae .
There has been a focus on internal antigens
The development of molecular biology technologies facilitated screening of bacterial (and later yeast) expression libraries with sera to select highly reactive clones for the production of recombinant proteins and their testing for vaccine potential in the mouse model. The pitfall of this approach is that abundant internal proteins are highly immunoreactive, precisely because they have not been subjected to immunological pressure in the live worm . In consequence, expression screening produced a long list of cytosolic and cytoskeletal proteins with supposed vaccine potential. WHO-TDR (ca. 1990) sponsored independent trials in mice for six candidates (three cytosolic, two cytoskeletal, and one transmembrane) , which failed to deliver adequate protection . Nevertheless, two of the cytosolic proteins have eventually proceeded to human vaccine trials [27,28] with an unknown outcome. A recent label-free quantitative proteomic analysis of the soluble proteins in an adult worm homogenate (soluble worm antigenic proteins, SWAP)  casts further doubt on the suitability of internal proteins as vaccines. Half the total mass comprised only 18 proteins, 11 of which (25% of total mass) had shown protective potential when used singly to vaccinate mice. Thus, SWAP appears to be a rich soup of supposed vaccine candidates. Why then is it poorly protective in mouse vaccine experiments, revealing an absence of synergy between candidates? The failure of SWAP due to the presence of inhibitory components is a weak argument; perhaps, the problem lies in the utility of the mouse model for vaccine testing .
Is there an intrinsic flaw in the mouse model that produces spurious protection data?
The mouse is a very convenient laboratory host in which to grow schistosomes, but is it optimal for vaccine testing? Migration from the skin is entirely intravascular (reviewed in ref. ). In a naive mouse, only ∼30% of penetrating cercariae actually mature into adult worms; in hamsters, the value is 56–76% and in baboons >80% . Passage through capillary beds is a tight squeeze for which larvae become very elongate with the lungs presenting the greatest obstacle to migration. The fragility of pulmonary capillaries, related to animal size, and their susceptibility to a cytokine-induced vascular leak syndrome is well documented . During lung transit in mice, some schistosomula burst into the alveolar spaces [33,34], yet possess only a limited capacity to re-enter tissues. This provides a physical explanation for the low level of maturation . Nevertheless, the mouse has been the favored animal for vaccine testing, despite the fact that a 50% reduction in worm burden translates to only a further 15 worms above the 70 out of 100 that would fail to mature anyway. The practice in mouse trials has increasingly been to shorten the interval between the last antigen boost and cercarial challenge (often 2–3 weeks and sometimes only 10 days). Our contention is that after such vaccination protocols, challenge parasites will reach the lungs when both activated T cells and cytokine levels from the preceding vaccination are maximal in the circulation . We have suggested that ‘protection’ in this situation is simply the result of physiological effects on the pulmonary blood vessels, increasing the proportion of parasites that enter the alveoli. The hypothesis provides an explanation for why internal antigens (including unlikely candidates such as nucleic acid-binding Y-box protein), plus a variety of heterologous proteins, reduce the level of maturation in a non-antigen-specific way. This argues for trials in more realistic larger hosts with less fragile pulmonary capillaries and a higher percentage maturation of penetrant cercariae, like the baboon (S. mansoni) and rabbit (S. japonicum).
Is the radiation-attenuated vaccine still a valid model?
The attenuation of cercariae by irradiation such that they undergo a truncated migration yet induce protection against a challenge with normal larvae has been used as a model vaccine for decades, mostly in the laboratory mouse (reviewed in refs [35,36]). Given the above critique of the mouse model, are protection data obtained in mice using the radiation-attenuated cercarial vaccine still valid? The crucial features in its favor are as follows:
The lung stage is the target of the effector mechanism, which acts via a focal inflammatory response  to trap or deflect additional parasites, beyond the normal fraction.
The trapped larvae are apparently not harmed, just blocked en route to the portal system .
From the mouse studies, it follows that antigens secreted by the lung stage are the likely targets in this model. Biosynthetic labeling of day 7 schistosomula found a dominant secreted product of ∼20 kDa , more recently identified by proteomics as a cocktail of MEG-3 proteins (Smps 138060, 138070 and 138080; ). Parenthetically, the same proteins are secreted by the mature egg. It is plausible that they may have a role in its escape mechanism from blood vessel to tissues by interacting with receptors on vascular endothelium . These lung-stage secretory proteins represent the best targets highlighted by the attenuated vaccine model and would repay testing for vaccine potential.
The composition of the parasite–host interface
The advent of proteomics and RNA-Seq has made possible the characterization of the adult tegument surface and the gut vomitus fractions of the host–parasite interface  (the composition of the gastrodermal surface remains elusive due to its inaccessibility ). Techniques, such as enzymatic shaving  and surface biotinylation [50,51] of live worms, have helped pinpoint proteins exposed on, or secreted from, the tegument surface. However, as reports have multiplied, another pitfall has emerged, namely the incautious handling of worms during perfusion and in vitro incubation. Consequently, inventories of gut and tegument surface proteins are now heavily contaminated by the familiar and highly abundant internal proteins (e.g. ). Of equal concern is the current focus on parasite-derived exosomes, and the description of their release at the tegument surface during in vitro culture . Such vesicles were never seen in the early TEM studies of adult worms dissected out and fixed immediately, without a perfusion step [16,54,55]. This is a strong argument that the exosome vesicles are an artifact of in vitro procedures.
What have we gleaned from these compositional studies? Taking a conservative view, adult schistosomes do not have a unique cell biology involving secretion of cytosolic and cytoskeletal proteins. Unless, a protein contains an N-terminal signal peptide, denoting export via the classical endoplasmic reticulum and Golgi pathway, it is best not included in secreted/surface-exposed categories unless complementary approaches, e.g. microscopy, reveal otherwise. For example, calpain and some annexins can be demonstrated at the tegument surface by immunocytochemistry [56,57], although their accessibility to antibodies in the live worm is presently unclear. In addition, the use of SecretomeP (http://www.cbs.dtu.dk/services/SecretomeP/) to justify inclusion should be avoided [58,59]. The appearance of histones in a list of putative secretory proteins predicted by the algorithm  ought to ring alarm bells. The schistosome genome encodes ∼1000 proteins with a conventional signal peptide, but schistosomes are also complex metazoan animals. Therefore, an additional criterion for inclusion in the external secretome has to be cytological evidence for expression/secretion at either tegument or alimentary tract interfaces with the host immune system, not at some internal location.
The capacity for a molecular definition of the parasite–host interface has produced some change in emphasis, with testing of tegumental proteins that fit the above criteria (TSP-2, Sm29, Sm200, Smp80, and calpain; [61–64]), as single vaccine candidates in the murine model. Both TSP-2 and calpain are being readied for Phase I trials in humans [65,66]. So far, only calpain has been successfully transferred to the baboon model with partial protection elicited , particularly effective against female worms. Persistence of high anti-calpain titers has also been demonstrated  although their relationship to protection is not known. The use of papain-type secreted gut cathepsins to elicit protection in the mouse model is also intriguing [69,70]. The cathepsin immunogens need to be enzymatically active, suggesting that a note of caution is required. The fact that papains from plants  and the liver fluke Fasciola  are also protective may indicate that the mechanism relies on a bystander effect, potentially involving mast cells and eosinophils, rather than specific acquired immunity .
Self-cure in the rhesus macaque as a route to a vaccine
A few laboratory hosts are able to prevent a patent schistosome infection from establishing, or to eliminate an existing one. Self-cure in the laboratory rat is one such, but can perhaps be ruled out as a vaccine paradigm due to a correlation with specific IgE production and hepatic mast cell infiltration , although the recent designation of nitric oxide as the potential ‘killer’ agent has added another dimension  to the mechanism of worm elimination.
Ironically, self-cure in the rhesus macaque appears an altogether more promising paradigm. We have revisited the model using both S. mansoni  and S. japonicum  infections, where the outcomes are strikingly similar. After exposure to 1000 cercariae, large numbers of adult worms establish (estimated 44–65% of penetrants at 8 weeks ) and oviposition begins at week 5. From ∼8–10 weeks, egg excretion in the feces tails off, followed later by a decline of worm gut-derived antigens in the circulation. Surviving worms recovered by perfusion at 18 weeks are emaciated, have stopped feeding, and are apparently starving to death [73,74]. Such macaques are resistant to a subsequent cercarial challenge, having acquired lasting protection .
So, what is our hypothesis about mechanism and mediating antigens? Initial observations on the fate of S. mansoni indicated that antibodies were elicited against both tegument surface proteins and gastrodermal secretions . The subsequent experiment with S. japonicum implicated the schistosome esophagus as a target for macaque antibodies  and led to an in-depth study of esophageal gland morphology and secretions (Figure 2; [76,77]). The esophagus is much more than a simple tube for conducting blood from mouth to gut. The lumen comprises two compartments, each surrounded by a gland from which proteins are secreted to initiate the processing of blood as it is ingested . A combination of RNA-Seq, in situ hybridization, and immunocytochemistry identified the major fraction of secreted proteins as encoded by microexon genes (MEGs), with a smaller complement of lysosomal hydrolases, and some potentially cytotoxic products (Table 1; ). These >40 gland products are proteins that schistosomes must deploy as part of their blood-feeding process. Furthermore, there is now abundant evidence from electron microscopy and immunocytochemistry that they can serve as in vivo targets for antibodies, which could potentially neutralize function [76,77]. As such, they represent an entirely new and untested category of vaccine targets within the context of acquired immunity in the rhesus macaque model that require investigation.
Confocal micrographs of the esophageal region of male S. japonicum.
|GeneDB #||Description||Log2 RPKM||GeneDB #||Description||Log2 RPKM|
|Smp_136830||Aspartyl protease||10.92||Smp_132470||Aspartyl protease||6.34|
|Smp_142970||P P thioesterase 1||10.43||Smp_205390||Aspartyl protease||6.14|
|GeneDB #||Description||Log2 RPKM||GeneDB #||Description||Log2 RPKM|
|Smp_136830||Aspartyl protease||10.92||Smp_132470||Aspartyl protease||6.34|
|Smp_142970||P P thioesterase 1||10.43||Smp_205390||Aspartyl protease||6.14|
Abbreviation: P P thioesterase 1, Palmitoyl Protein thioesterase 1.
Data modified from ref. 
RPKM is a normalized value for transcript abundance. Genes are listed by a decreasing RPKM value.
Described by Almeida et al.,  but not annotated in GeneDB.
Gene removed from GeneDB but abundant transcripts in male heads.
The single antigen magic bullet versus multiple antigenic targets
From many angles, the discovery of a single antigen that conferred a high level of protection against schistosome infection would be ideal and, although not articulated, that is what virtually all experiments with recombinant proteins have been aiming for. Is this realistic? Schistosomes are not hepatitis B virus with its major antigenic surface protein . The complexity of the schistosome genome (∼11 000 genes) with built-in redundancy of function makes single-target vulnerability improbable. We are dealing with a parasite which can adapt its own immune evasion responses to host attack by up-regulating its antioxidant systems as it develops into the adult worm [79,80]. A multiplicity of targets seems a more likely scenario in line with the slow attrition inflicted by the immune system of the rhesus macaque on its resident worm population (or the blocked migration of schistosomula in the lungs in the attenuated vaccine model). Once the primary self-cure response of the rhesus macaque is established, it may persist because the targets are expressed by challenge worms over the 3–4 weeks of their development in the portal system, ample time to trigger a memory response.
The task of identifying multiple antibody targets might seem daunting, but initial screening of arrays comprising protein  and glycan  targets with serum from self-cured rhesus macaques has already been undertaken. The real task is to decide what to include in such array screens; for the proteins that means, keeping in mind, the pitfall of uncritical proteomic analysis of parasite fractions. There is a need to focus on the crucial exposed/secreted targets within the huge background noise provided by highly reactive cytosolic and cytoskeletal proteins. Other approaches to identify conformational epitopes, which require correct protein folding, include screening phage display transcriptome libraries  with self-cure serum and the harder slog of expressing tegument membrane proteins in eukaryotic systems before screening the products by western blotting using infection sera .
Conclusions — what are the prospects?
More than 40 years of effort has not led to an effective schistosome vaccine. This should give pause for reflection, both about the underlying concepts and the focus on internal immunogenic proteins. This lack of progress has undoubtedly contributed to the prevailing skepticism about schistosome vaccine development. However, there are avenues based on recent research that have yet to be explored.
Important lessons can be learned from the rhesus macaque's ability to eliminate adult worms, where the in vivo reality seems to be death by slow starvation, mediated by antibodies directed against exposed or secreted proteins. The cohort of ∼40 esophageal proteins comprises novel and untried candidates.
There are new techniques, including arrays, to screen sera for reactivity with exposed/secreted proteins and identify immune correlates of self-cure status.
It is likely that potential targets need to be administered collectively to suitable hosts using formulations that elicit responses capable of degrading worm physiology to the point of organ failure and death.
Persistence of elicited responses is the key, with the potential for triggering immunological memory upon subsequent exposure. Using systems biology to establish the immune system signature for self-cure in the rhesus macaque would greatly aid vaccine formulation to this end.
Self-cure from a schistosome infection by the rhesus macaque provides the most promising context for vaccine development.
Proteomic and RNA-Seq approaches, combined with studies on the esophageal glands, have defined a cohort of proteins secreted from or exposed at the schistosome–host interface, which may provide the targets for the self-cure process.
Identification of the most reactive and their formulation in modern adjuvants should trigger a new round of vaccine testing in realistic model hosts.
The Authors declare that there are no competing interests associated with the manuscript.