Screening of a Library of Recombinant Schistosoma mansoni Proteins With Sera From Murine and Human Controlled Infections Identifies Early Serological Markers

Abstract Background Schistosomiasis is a major global health problem caused by blood-dwelling parasitic worms, which is currently tackled primarily by mass administration of the drug praziquantel. Appropriate drug treatment strategies are informed by diagnostics that establish the prevalence and intensity of infection, which, in regions of low transmission, should be highly sensitive. Methods To identify sensitive new serological markers of Schistosoma mansoni infections, we have compiled a recombinant protein library of parasite cell-surface and secreted proteins expressed in mammalian cells. Results Together with a time series of sera samples from volunteers experimentally infected with a defined number of male parasites, we probed this protein library to identify several markers that can detect primary infections with as low as 10 parasites and as early as 5 weeks postinfection. Conclusions These new markers could be further explored as valuable tools to detect ongoing and previous S mansoni infections, including in endemic regions where transmission is low.

Schistosomiasis is a neglected tropical disease affecting more than 200 million people in 52 countries and is one of the world's major health problems causing 200 000 deaths per year. In 2015, the impact of the disease was estimated at 3.5 million disabilityadjusted life years, putting a huge socioeconomic burden on many low-and middle-income countries [1]. In humans, schistosomiasis is caused by 5 species of platyhelminth parasites belonging to the genus Schistosoma. Their geographical distribution is restricted by the presence of species-specific freshwater snails that act as intermediate hosts, with Schistosoma mansoni having the most widespread distribution, encompassing both Africa and South America [2]. Infected snails shed cercariae, a free-swimming larva, which penetrates human skin to initiate infection. Parasite maturation within its host takes several weeks: cercarial heads first remodel their surface to form schistosomula [3], which migrate through the dermis for several days before entering the host bloodstream. Eventually, young adult male and female S mansoni worms pair up in the liver before moving to the mesenteric blood vessels, where each pair can release over 300 eggs per day from 5 weeks postinfection [4]. The symptoms of schistosomiasis are caused by progressive accumulation of eggs within host tissues, eliciting host-derived immune responses that can eventually lead to liver fibrosis, portal hypertension, and, if left untreated, death [5]. In the absence of a licenced vaccine, treatment relies on the use of a single drug, praziquantel.
The exact prevalence of schistosomiasis worldwide may be underestimated due to limitations of routine methods of detection [6]. Diagnosis of S mansoni infections and subsequent decisions on mass drug administration mainly rely on observing parasite eggs in patients' stools with the Kato-Katz test. Although this method can detect current infections, it is not sensitive enough to diagnose low levels of infection present in low-endemicity areas or in recently treated populations [7]. In these instances, detection of parasite antigens such as the circulating cathodic antigen or circulating anodic antigen (CAA) in patients' sera or urine by sensitive field-applicable or laboratory-based assays is highly useful [8][9][10]. In areas where elimination has been achieved, the detection of antiparasite serum antibodies could support epidemiological surveillance that informs of any risk of resurgence. Host antibody responses to S mansoni parasites is often measured against whole parasite extracts such as soluble egg antigens (SEA) or soluble worm antigen preparation (SWAP), which are not molecularly defined and can lead to cross-reactivity with other helminth species. Although a diagnostic test based on a recombinant protein could mitigate these problems, only very few have been used to test patient antibody responses to Schistosoma infections [11]. The development of new, more sensitive diagnostic tools would therefore help to improve the detection and early treatment of schistosomiasis. Extracellular antigens released by or displayed at the surface of the parasite at the initial stage of infection [3] can be valuable early immunodiagnostic markers because they are directly exposed to the host humoral system. The identification of such antigens has been aided by the sequencing and annotation of the S mansoni genome [12], and several proteomics [13][14][15][16][17][18][19][20], transcriptomics, and in silico analyses [21][22][23] have identified genes expressed by the schistosomula and adult worm. Despite their value, extracellular proteins pose challenges for recombinant expression because they contain structurally important posttranslational modifications such as glycosylation and especially disulphide bonds required to produce informative antibody epitopes. To address this, we have previously developed protein expression approaches in mammalian cells to compile large panels of parasite recombinant ectodomains that retain their binding activity and immunogenicity [24], enabling the identification of host-parasite receptor-ligand interactions and humoral markers of protection against malaria [25].
To identify new markers for S mansoni infections, we created a panel of 115 recombinant proteins representing secreted and membrane-tethered S mansoni proteins mostly enriched at the schistosomula stage. Using human and mouse sera from experimentally controlled S mansoni infections, we determined the kinetics of humoral responses to this panel of antigens in the context of primary infections and identified several early serological markers.

Identification of Schistosoma mansoni Cell-Surface and Secreted Proteins
Genes encoding cell-surface and secreted proteins from cercarial and adult S mansoni were identified using published proteomic and transcriptional data [13][14][15][16][19][20][21]. To further enrich for genes transcribed at the schistosomula stage, we identified 1302 transcripts upregulated at 48 versus 3 hours posttransformation [26] using EdgeR [27] as well as the 1000 most abundant transcripts at 48 hours as assessed by reads per kilobase mapped, resulting in 1977 unique genes, 274 of which encoded secreted and cell-surface proteins identified using signal peptide and transmembrane domain prediction software [28,29]. All RNA-seq data are available in the European Nucleotide Archive (ENA) under study PRJEB3190 and run accession numbers ERR411525, ERR411535, and ERR411541 for the 3-hour time point and ERR411522, ERR411527, and ERR411538 for the 48-hour time point. Ninety-four mitochondrial, endoplasmic reticulum, or multipass proteins, which are difficult to express as a contiguous ectodomain, were subsequently excluded, leaving a short list of 180 proteins. Gene structures were manually refined by mapping transcriptome data to the genome sequence, and genes spanning gaps in the genome sequence or with ambiguity in structure were removed, resulting in a final list of 115 candidates numbered in Table 1.

Recombinant Protein Expression Using a Mammalian Expression System
The entire ectodomain of membrane-anchored proteins was selected, their signal peptide removed using predictions from SignalP v3.0, and the corresponding cDNAs were made by gene synthesis after codon-optimization for expression in human cells (GeneArt; Invitrogen). The ectodomains were flanked by unique NotI and AscI restriction sites and subcloned into an expression plasmid containing the mouse V K 7-33 signal peptide [24], and a C-terminal tag containing rat Cd4d3 + 4 domain, a BirA monobiotinylation sequence, and 6-His tag (Addgene plasmid no. 50 803) [30]. All expression constructs are available at www.addgene.org (plasmids nos. 120 590 to 120 704). Plasmids were transiently cotransfected with BirA in HEK293-E or -6E cells, supernatants were collected, and recombinant proteins were detected by Western blotting with 0.02 µg/mL streptavidin-HRP (Jackson ImmunoResearch) as previously described [24]. When proteins showed signs of proteolytic cleavages, transfections were repeated in the presence of a protease inhibitor cocktail (Sigma).

Infections
To initially characterize the immunoreactivity of the proteins, we used plasma pools from 10 nonexposed European controls and 10 Ugandan adults from a cohort living in a high transmission area [31,32]. Individuals were selected from those who had a soluble worm antigen immunoglobulin (Ig)G1 response in the upper third for the cohort with a median egg count of 876 eggs per gram (epg) (interquartile range, 345-1967 epg). All patient samples were collected in accordance with the Uganda National Council for Science and Technology and the Cambridge Local Research Ethics Committee. Sera or plasma from volunteers experimentally infected percutaneously with 10-30 male cercariae were collected weekly in accordance with the LUMC Institutional Medical Ethical Research Committee (P16.111), as previously described [33]. All volunteers were treated with praziquantel 12 weeks after infection.

Sera From Experimental Mouse Infections
The life cycle of the NMRI (Puerto Rican) strain of S mansoni was maintained by routine infections of mice and susceptible Biomphalaria glabrata snails under the UK Home Office Project Licence nos. P77E8A062 and PD3DA8D1F; all protocols were

Enzyme-Linked Immunosorbent Assays
Protein expression was quantified by enzyme-linked immunosorbent assay (ELISA) as previously described [24]. Briefly, serial dilutions of biotinylated proteins were captured on streptavidin-coated microtitre plates and detected by mouse antirat Cd4 OX68 antibody (AbD Serotec), followed by an alkaline-phosphatase-conjugated antimouse secondary antibody (Sigma), and proteins were classified into high (typically >5 µg/mL transfection supernatant), medium (between 1 and 5 µg/mL), and low (<1 µg/mL) levels of expression. To determine the presence of heat-labile epitopes, biotinylated proteins were captured on streptavidin-coated plates either untreated or after heat treatment for 10 minutes at 80°C before incubation with sera at 1:1000 dilution in HBST/2% bovine serum albumin (HBST/2%BSA). Statistical analysis was performed in GraphPad Prism using the Holm-Sidak method for multiple t tests. Sera from human volunteers or experimentally infected mice were diluted 1:250 or 1:1000, respectively, in HBST/2%BSA; binding was detected with horseradish peroxidase-conjugated antihuman or antimouse secondary antibodies (Sigma), respectively, recognizing IgA, IgM, and IgG. For each individual, the optical density (OD) value of the preinfection samples was reference-subtracted from the OD readings at all subsequent time points. The OD values for each protein were compared with that of the rat Cd4d3 + 4 tag used as a negative control, and seropositivity was defined as OD protein > OD control + 3SD control .

Schistosoma mansoni Recombinant Proteins Expressed in Mammalian Cells Contain Heat-Labile Conformational Epitopes
Many antibodies elicited in the context of a natural infection recognize conformational epitopes present on the native protein.
To determine the fraction of proteins containing conformational epitopes, we compared the immunoreactivity of pooled plasma from chronically exposed individuals to untreated and heat-treated proteins. All except proteins 18 and 57 were seropositive, and 66 of the expressed proteins (64%) were highly immunoreactive (A 280 > 0.3) (Figure 2). The majority of proteins showed moderate to strong loss of immunoreactivity after heat treatment; only 16 of the highly reactive proteins showed no statistically significant loss of reactivity, suggesting that they were either natively unstructured, misfolded, or contained heat-stable domains ( Figure 2). Overall, the recombinant proteins were therefore immunoreactive to plasma from individuals living in high-endemicity areas and contained heat-labile epitopes.

Infections
The recent establishment of controlled human S mansoni infections [33,34] provided a unique opportunity to identify early serological markers of infection and the kinetics of the host antibody response in an experimentally controlled setting. We initially tested plasma taken at 4 weekly intervals from 3 volunteers, each infected with 30 male cercariae, against our 103 expressed S mansoni proteins. As expected, compared with high-endemicity plasma, fewer antigens were immunoreactive, but several were immunopositive in all participants, and the number of positive antigens increased over time (Figure 3).   Figure 2. The majority of recombinant proteins are immunoreactive to sera from individuals living in schistosomiasis-endemic areas and contain heat-labile conformational epitopes. Recombinant proteins were probed with pooled sera from individuals living in schistosomiasis-endemic areas ("infected," red checked bars) or individuals from the United Kingdom who have never been infected ("control," green dotted bars). To test for the presence of heat-labile epitopes, recombinant proteins were also heat-treated (80°C, 10 minutes) before being exposed to immune sera ("infected heat treated," orange hatched bars). All except 2 proteins (18 and 57, shown in blue) were seropositive, as determined by A protein > A control + 3SD control ( = 0.201) (red dashed line), where control is the rat Cd4d3 + 4 protein tag. High immunoreactivity was determined as A protein > 0.3 (green dotted line). Proteins that exhibited little or no loss of reactivity after heat treatment are shown in italics, including 16 highly reactive proteins (6,7,14,16,19,30,32,34,36,42,49,54,56,89,101,105). All measurements were performed in triplicate; error bars = standard deviation.
Early diagnosis of infection before the onset of symptoms would be very valuable; therefore, we analyzed weekly samples from all 3 individuals until week 8 against the 9 antigens positive in all volunteers at the 12-week time point (Figure 4). Immunoreactivity was observed in all volunteers as early as 5 weeks postinfection for proteins 44 and 65, 6 weeks for protein 68, and 7 weeks for proteins 63 and 67. As observed previously, only 2 of the 3 participants were reactive to proteins 61, 62, 83, and 106 between 5 and 8 weeks postinfection. In the case of antigens 44, 65, 68, and 106, seropositivity increased sharply between weeks 4 and 6 before plateauing at later time points.

Broadly Similar to That of Humans
Although sera from experimental infections with male carcariae identified valuable diagnostic antigens, any female-restricted antigens would not have been identified. To address this in a controlled experimental setting, we used mice as an animal model of infection. Sera were collected from mice infected either percutaneously with 200 cercariae or through intraperitoneal injection of 350 cercariae, and their serological responses were quantified ( Figure 6). None of the antigens showed consistent reactivity across samples at 8 days postinfection, as expected; however, reactivity to antigen 44, and to a lesser extent, antigen 3, could be detected across all samples from day 21. At 42 days, an additional 8 antigens showed reactivity across all sera tested. Overall, reactivity with the pool of sera from mice infected intraperitoneally was stronger than individual mice infected percutaneously, possibly due to the higher number of cercariae used. Just like in human samples, proteins 44, 62, 63, 67, 68, and 106 were the most immunoreactive. In addition, reactivity to the Ly6 family members Ly6F, Ly6B, and Ly6D (proteins 3, 10, and 41, respectively) was detected at an earlier time point in mice than it was in humans. It is interesting to note that proteins 61 and 65, which were positive between 5 and 6 weeks postinfection in humans, did not react with murine samples, whereas conversely protein 71 generated a stronger response in mice.

DISCUSSION
Despite its widespread distribution and high morbidity rate, schistosomiasis remains a neglected tropical disease whose true incidence and health impact remains underestimated [6].
In this study, we have described a recombinant protein library containing secreted and cell-surface proteins from different developmental stages of S mansoni, and together with sera obtained from a recently established controlled human infection model [33] we used it to determine the kinetics of the humoral response to S mansoni infection. Although other large arrays of parasite proteins have been used for diagnostic and immuno-epidemiological purposes, they have mostly relied on cell-free or bacterial expression systems [35,36]. Mammalian expression systems are more suitable for the addition of structurally important posttranslational modifications found on extracellular proteins and thereby preserve conformational epitopes that can be recognized by antibodies. Using this approach, we successfully expressed 103 proteins, the majority of which were observed at their expected size. Although the 12 proteins we could not express were equally distributed among the different protein families represented in the library, 3 of them corresponded to elastases. Using pooled plasma from patients living in a high-transmission region of Uganda, we observed that 64% of the expressed proteins were strongly immunoreactive (A 280 > 0.3) with the majority (82%) showing sensitivity to heat treatment, an indicator of tertiary folding.
Although the lower immunoreactivity for 37 proteins may suggest incorrect folding, they could also be weakly immunogenic in humans or not directly exposed to the host immune system. Of the 16 most immunoreactive proteins observed, 11 had already been identified in the surface and secreted proteome of S  mansoni; by contrast, only 6 of the 37 proteins with low serum reactivity (16%) have already been described in proteomics studies. The use of a single mammalian expression system for the production of large panels of proteins is particularly attractive for the systematic comparison of antigens in diagnostic, immuno-epidemiological, or vaccination studies, and we have used this approach previously to identify potentially protective antibodies against malaria [24,25,30].
Mass administration of praziquantel to schoolchildren has been the mainstay of control programs against schistosomiasis, and this has proved relatively successful in reducing parasite burdens and contributing to elimination from some areas [37]. Continued surveillance and early detection of new cases remains critical to avoid any risk of resurgence; however, the commonly used Kato-Katz method is not sensitive enough to detect low levels of parasitemia, resulting in the underestimation of the real number of cases [7]. Detection of parasite-derived glycans such as CAA in the urine or serum of patients by lateral-flow test is currently considered the most sensitive assay for the detection of current Schistosoma infections because it can detect very low infection levels and reactivity disappears rapidly after praziquantel treatment [8,38]. In areas where schistosomiasis has been eliminated or is close to elimination, alternative methods of surveillance might be needed. By persisting several months or even years after infection [39], antibody responses provide a useful historical measure of parasite exposure to monitor populations at risk of resurgence. Currently, most host antibody responses are measured against crude parasite preparations such SEA or SWAP, which may suffer from considerable cross-reactivity with other helminths antigens. Therefore, species-specific recombinant proteins as diagnostic tools could be more reliable.
A striking feature of this study is the relatively small number of antigens eliciting a patent immune response in the few weeks after a primary infection. In humans, almost no IgM/IgG response could be detected before 4 to 5 weeks postinfection: some antigens might be more highly expressed at later stages of parasite development, or the parasite could evade the host immune response. Subsequently, antibody responses were almost exclusively directed at secreted proteins belonging to the saposin and cathepsin families. Of the 9 proteins reactive in all human volunteers before 12 weeks, 5 contained saposin domains and 3 were cathepsins. It is interesting to note that several saposin domain-containing proteins from Schistosoma japonicum have been proposed as markers of infections in mice and humans [40][41][42], whereas proteins 66 and 67, 2 isoforms of the cathepsin B1 protease, seem to have built-in adjuvanticity [43]. Both saposins and cathepsins are produced by the schistosome's alimentary tract and have been involved in the digestion of lipids and proteins [44,45]. Their reactivity to the host humoral system suggests they are produced by the parasite at an early stage. Although 5 of the 6 saposin-domain-containing proteins  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  28  29  30  31  32  33  34  35  36  37  38  40  41  42  43  44  45  46  47  48  49  51  52  53  54  55  56  57  58  59  61  62  present in the library were immunoreactive (although reactivity of protein 83 remained modest up to 12 weeks), they share only 9% to 28% sequence identity, making cross-reactivity unlikely. The dynamics of the host response to the recombinant saposins closely parallels the detection of the parasite CAA glycans in the sera of infected patients, which is first detected at 4 weeks [38]. Reactivity to members of the uPAR/Ly6 domain-containing family, which are expressed at the somule and adult stages [46][47][48], was also consistently observed across human and murine samples. In our study, reactivity to Ly6F, Ly6B, and Ly6D (proteins 3, 10, and 41, respectively) was observed in both human and mouse serum samples, whereas reactivity to Ly6A and Ly6I (proteins 11 and 2, respectively) was only detected in human samples. Three of these proteins (Ly6B, Ly6D, and Ly6F) have been shown to elicit strong antibody responses in rat, mouse, and human sera [48]. Although they share some homology with the complement-inactivating Cd59 protein, they do not seem to have conserved the same function [46].

CONCLUSIONS
By producing recombinant proteins in a mammalian expression system, we have paid particular attention to their correct folding, and thus this new resource could be used in a wide range of cellular and molecular assays such as vaccine screening [49], cellular assays looking at immunomodulatory functions, immunoepidemiological studies, or the identification of host binding partners by receptor-ligand screening. We envisage these proteins will be useful to the wider scientific community to further understand Schistosoma biology.