A rotifer-derived paralytic compound prevents transmission of schistosomiasis to a mammalian host

Schistosomes are parasitic flatworms that infect over 200 million people, causing the neglected tropical disease, schistosomiasis. A single drug, praziquantel, is used to treat schistosome infection. Limitations in mass drug administration programs and the emergence of schistosomiasis in nontropical areas indicate the need for new strategies to prevent infection. It has been known for several decades that rotifers colonizing the schistosome's snail intermediate host produce a water-soluble factor that paralyzes cercariae, the life cycle stage infecting humans. In spite of its potential for preventing infection, the nature of this factor has remained obscure. Here, we report the purification and chemical characterization of Schistosome Paralysis Factor (SPF), a novel tetracyclic alkaloid produced by the rotifer Rotaria rotatoria. We show that this compound paralyzes schistosome cercariae and prevents infection and does so more effectively than analogous compounds. This molecule provides new directions for understanding cercariae motility and new strategies for preventing schistosome infection.


Introduction
Schistosomiasis-caused by parasitic flatworms of the genus Schistosoma-is a major neglected tropical disease, affecting over 200 million people, with over 700 million people at risk of infection [1][2][3]. Praziquantel is currently the only drug used for treating schistosomiasis. Concerns about the emergence of drug resistance [4,5], as well as limitations observed in mass drug administration programs [6][7][8][9], highlight the need to devise new strategies for preventing infection by these parasites. This need is amplified by the recent identification of people infected with human-livestock hybrid schistosomes and the geographical expansion of schistosomiasis to temperate regions [10][11][12].
Schistosomes have a complex life cycle that alternates between an intermediate host (snail) and a definitive host (mammal) via 2 free-living, water-borne forms called miracidia and cercariae, respectively [13] (Fig 1A). For decades, inconsistency in cercarial production by snails and infectivity of mammalian hosts has been observed in most schistosome laboratories [14]. Intriguingly, Stirewalt and Lewis reported that rotifer colonization on shells of the snail intermediate host (Biomphalaria glabrata) significantly reduced cercariae output, motility, and infectivity [15]. Furthermore, they observed that cercarial motility was affected not only by the presence of rotifers but also by rotifer-conditioned water, indicating that rotifers released water-soluble molecules with paralytic activity. Almost 40 years have passed since this important finding, yet this factor's identity has remained a mystery.

Purification of the rotifer-derived compound
Encouraged by this anticercarial effect and its potential to prevent schistosome infection, we sought to purify this paralyzing agent. We isolated individual rotifers from snail shells and found 2 species, R. rotatoria ( Fig 1B) and Philodina acuticornis (Fig 1C), as previously reported by Stirewalt and Lewis [15]. To identify which rotifer was responsible for the paralytic effect, we grew clonal isolates of each species, producing rotifer-conditioned artificial pond water (APW). Adding R. rotatoria-conditioned APW to freshly collected cercariae resulted in gradual paralysis within 5 min (Fig 1D). Most cercariae stopped swimming and sank to the bottom of the dish. Tapping the dish could stimulate their movement, but their response was limited to writhing on the dish bottom or short-distance swimming before becoming paralyzed again. In contrast, P. acuticornis-conditioned water had no effect ( Fig 1E).
To purify the paralyzing agent, we performed molecular weight cut-off (MWCO) filtration of rotifer-conditioned water and found that the activity was present in the <650 Da fraction. The <650 Da filtrate was fractionated by reversed-phase high-performance liquid chromatography (RP-HPLC; Fig 2A), and each fraction was tested on cercariae. Paralysis was only observed following treatment with a peak eluting at 25 to 27 min ( Fig 2B). As expected, this peak was detected only in R. rotatoria-but not P. acuticornis-conditioned water ( Fig 2B). A second round of HPLC on this peak revealed one peak (eluting at 24-26 min) with paralytic activity (Fig 2C). A predominant signal of 273.16 Da (protonated molecular ion [M+H]) in this peak was revealed by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS; Fig 2D). Consistent with the paralysis assay, this signal (m/z 273.16) was detected exclusively in the fraction eluting at 24 to 26 min but not in the fractions before or after ( Fig 2E). These results suggested that the component with m/z 273.16 was the paralyzing agent, which we named "Schistosome Paralysis Factor" (SPF). We then determined the monoisotopic mass for protonated SPF using high-resolution quadrupole time-of-flight (Q-TOF) MS, 273.1595 Da (Fig 2F), suggesting C 16 H 20 N 2 O 2 as the best-fitting formula for SPF.

SPF is a novel tetracyclic alkaloid
To elucidate its structure, we purified approximately 0.1 mg SPF from 25 L R. rotatoria-conditioned water. Nuclear magnetic resonance (NMR) spectroscopy revealed a novel tetracyclic structure. Briefly, 1 H spectra showed the presence of 19 protons in the compound (S1 Total correlation spectrometry (TOCSY) showed that aliphatic protons, except 2 methyl groups, are from one spin system (S4 Fig). The connectivity of the neighboring groups was derived from correlation spectroscopy (COSY) and heteronuclear multiple-bond correlation (HMBC) spectra (S5 Fig and S6 Fig). Overall, the aliphatic region is composed of a dimethylpyrrolidine structure, which is linked to an indole via a CH 2 group and an oxygen. Nuclear Overhauser effect spectroscopy (NOESY) suggested (R, S, S) or (S, R, R) configurations on the chiral centers (S7 Fig). Altogether, combined NMR analysis led to 2 possible structures ( Fig 2G  and 2H and S1 Table).

SPF and its analogs paralyze cercariae in a dose-dependent manner
To test its dose dependency, we examined the paralytic effect of serially diluted SPF on cercariae by quantifying their movement over time. In the absence of SPF, over 82% of cercariae were free swimming over 3 min (Fig 3A and S1 Data). In 2.5 nM SPF, the percentage of freeswimming cercariae dropped to 67% 3 minutes after drug treatment. As the concentration of SPF increased, so did the rate of paralysis, and more cercariae were paralyzed at the end of treatment. We observed maximum effects in 250 nM and 2.5 μM SPF, with the majority of cercariae paralyzed within 30 s.
Two natural compounds from Streptomyces sp., ht-13-A and ht-13-B, are structurally related to SPF; they were isolated based on their affinities for human serotonin receptors [17]. All 3 alkaloids share a novel oxepineindole framework fused with a pyrrolidine ring (Fig 3A-Fig 1. R. rotatoria-conditioned water paralyzes Schistosoma mansoni cercariae. (A) Life cycle of S. mansoni. Adult parasites, residing in the mammalian host vasculature, lay eggs (not shown). Upon exposure to fresh water, eggs release miracidia, which infect the appropriate snail host. Inside the snail, the parasite reproduces asexually, ultimately producing large numbers of free-swimming infective larvae (cercariae) that can penetrate mammalian skin to continue the life cycle (adapted from [16]). (B and C) Nomarski differential interference contrast microscopy images of R. rotatoria and Philodina acuticornis (arrowhead indicates the rostrum in R. rotatoria, which is lacking in P. acuticornis). Scale bars: 100 μm. (D and E) Maximum intensity projection (5 s, 150 frames) of cercariae motility after treatment with R. rotatoria-or P. acuticornis-conditioned water.
https://doi.org/10.1371/journal.pbio.3000485.g001 3C and S1 Data; note the serotonin backbone highlighted in red in Fig 3A). Although synthesis of SPF has not been achieved, total syntheses of ht-13-A and ht-13-B have been reported [18][19][20]. To test whether this shared tetracyclic scaffold is responsible for the paralytic effect, we analyzed structure-activity relationships by using ht-13-A, ht-13-B, 3 ht-13-A derivatives [18], and one epimer in cercarial paralysis assays. Importantly, ht-13-A, although not as potent as SPF, also had a paralytic effect on cercariae (Fig 3B and S1 Data). In contrast, ht-13-B did not paralyze cercariae, suggesting that the extra methyl group disrupts interaction with the target it was more potent than ht-13-A, indicating that the nature of the side chain is important for proper target interaction ( Fig 3D and 3E and S1 Data). In contrast to ht-13-A and Ht-13-A-pr, the epimer was unable to paralyze cercariae; these results support the (R, S) configuration of SPF at C12, 13 as the biologically active form ( Fig 2G).

SPF prevents mammalian infection
Because motility of the cercarial tail is essential for swimming and provides force for skin penetration [21][22][23], we examined whether SPF prevented infection. We treated approximately 200 cercariae with different concentrations of SPF for 10 min and then tested their infectivity by exposing them to mouse tails for 30 min (N = 6 for each condition). Six weeks post infection, we euthanized the mice, counted schistosomes recovered after hepatic portal vein perfusion, and examined liver pathology. From controls, we recovered 83 adult worms on average ( Fig 4B and S2 Data), consistent with typical recoveries of approximately 40% [24]. Livers from these mice appeared dark and contained extensive granulomas ( Fig 4A). In contrast, we did not recover any adult worms from mice after treatment with 250 nM or 2.5 μM SPF ( Fig  4B and S2 Data), and no granulomas were observed ( Fig 4A). Histological examination confirmed that these livers were free of schistosome eggs (  and/or chemical stimuli from mouse tails may overcome SPF-induced paralytic effects at low SPF concentrations. Notably, neither Ht-13-A nor Ht-13-A-pr blocked infection as completely from mice (N = 6) exposed to drug-treated cercariae. Livers from mice treated with control and lower drug concentrations were darker in color and contained more granulomas (white spots). With higher drug concentrations, livers had normal morphologies with few or no granulomas; 25 μM SPF treatment was ND because of limited amounts of purified SPF. (B-D) Numbers of adult worms recovered from exposed mice (2 experiments for each drug, 6 mice total for each condition). (E-G) Numbers of schistosome eggs per area (/mm 2 ) from liver sections (4-6 sections per mouse). Data for panels B-G are mean ± SD. Statistics: One-way ANOVA, post Dunnett's test. See S2 Data for corresponding raw data. ND, not determined; SPF, Schistosome Paralysis Factor.
https://doi.org/10.1371/journal.pbio.3000485.g004 as 250 nM SPF, even at 25 μM (Fig 4A, 4C, 4D, 4F and 4G and S2 Data). Under more realistic infection conditions, in which mouse tails were lifted 1 to 2 cm from the bottom of the test tube containing cercariae, so they had to swim actively towards the tail to infect the mouse, Ht-13-A and Ht-13-A-pr were still not as effective as SPF, which completely blocked infection (S8 Fig).

Conclusion
This work has identified a novel tetracyclic alkaloid, produced by the rotifer R. rotatoria, that paralyzes the infective larvae of schistosomes. Although its mechanism of action remains unknown, its chemical structure provides important clues. SPF contains a serotonin backbone, suggesting that SPF might antagonize serotonin signaling, perhaps via G protein-coupled receptors (GPCRs) or serotonin-gated channels. Consistent with this idea, the structurally related compounds, ht-13-A and ht-13-B, bind several human serotonin receptors [17]. In schistosomes, serotonin has been implicated in neuromuscular functions in multiple life cycle stages [25][26][27][28]; knocking down a serotonergic GPCR (Sm5HTR) in schistosomulae and adult worms led to decreased movement [29]. Interestingly, praziquantel partially activates the human serotonin receptor, HT2BR, suggesting that it may also target schistosome serotonergic GPCRs [30].
The chemical ecology underlying R. rotatoria's production of SPF is also unclear. Whether SPF is used naturally to combat other aquatic creatures (e.g., to prevent other rotifers from colonizing areas where R. rotaria live) and, thus, the effect on schistosome cercariae is indirect, or whether SPF benefits the rotifer's commensal host will require further study. Because compounds with structural similarities to SPF are produced by Streptomyces sp., it will be important to examine the possibility that SPF is not directly produced by the rotifer but rather by constituent(s) of its own microbiome. However, given that horizontal gene transfer is well documented in rotifers [31,32], it is also possible that R. rotatoria has acquired the synthetic machinery to produce SPF on its own. Future work will help reveal the source of SPF and its biosynthetic pathway.
In the past few decades, the discovery and development of natural products have helped combat parasitic diseases [33]. Based on its ability to block infection, SPF holds great promise as an antischistosomal agent. Identifying the biologically active chemical scaffolds and understanding SPF's mode of action are expected to provide important clues for preventing schistosomiasis. . To obtain S. mansoni cercariae, B. glabrata snails were exposed to light at 26˚C for 1 to 2 h. APW containing cercariae was passed through a 100 μm cell strainer (Falcon; Corning, NY) to remove snail food and feces. Cercariae were then collected using custom-made 20 μm cell strainers.

Rotifer culture
Because both rotifer species reproduce parthenogenetically, we clonally expanded each species into 1 L cultures from a single rotifer. Individual rotifers (R. rotatoria and P. acuticornis) were initially isolated from the shell of B. glabrata and cultured in APW in 24-well plates. Each individual colony was expanded into ever-larger culture volumes and ultimately maintained in 2 L flasks. Both species were fed Roti-rich liquid invertebrate food (Florida Aqua Farms Inc.; Dade City, FL). Rotifer-conditioned water was collected every month by filtering out the rotifers using a 20-μm cell strainer. Filtered rotifers were then passaged to fresh APW to propagate the cultures.

Crude rotifer-conditioned water preparation
One-liter rotifer media was lyophilized, reconstituted with 50 mL dH 2 O, and filtered through 10,000 and 650 MWCO Pall Minimate TFF Capsules with Omega membrane (Ann Arbor, MI). Filtrate (<650 Da) was freeze dried. For RP-HPLC, 300 mg of the dried material was dissolved in dH 2 O and run on a RP-HPLC-Merck Chromolith semi-prep RP-18e column (Darmstadt, Germany) at 5 ml/min using a gradient of 100% A (water) to 60% B (acetonitrile; ACN) in 60 min. A total of 10 mL fractions were collected and assayed for biological activity. Fractions containing biological activity were saved for further study.

Further purification of rotifer media
The bioactive fractions were pooled, freeze dried with SpeedVac (Savant, MA), reconstituted with 500 μL dH 2 O, and injected into a 4.6 mm diameter × 25 cm Symmetry column (Waters; Millford, MA). A Breeze2 analytical LC system (Waters; Millford, MA) was employed for separation at 0.5 mL/min with the following solvents and gradients: Solvent A, 0.1% formic acid (FA); solvent B, methanol with 0.1% FA; 0 to 10 min 0% to 10% B, 10 to 30 min 10% to 35% B, 30 to 33 min 35% to 80% B, 33 to 37 min 80% to 80% B, 37 to 40 min 80% to 0% B. Eluents were collected manually based on peak elution. All fractions were lyophilized, reconstituted with water, and analyzed with MALDI-MS. Fractions containing biological activity were saved for future use.

MALDI-MS analysis
For each collected fraction, 1 μL of sample solution was spotted on a ground steel MALDI target and mixed with 1 μL of alpha-cyano-4-hydroxy-cinnamic acid (CHCA; Sigma-Aldrich; St. Louis, MO) solution (10 mg/mL CHCA in 50% ACN solution with 0.005% trifluoroacetic acid). Mass calibration, spectra acquisition, and analysis were performed under conditions as previously described by Tillmaand and colleagues [35].

Hydrogen/deuterium exchange analysis
Acidified deuterated methanol (CD 3 OD, methanol-d4, Sigma-Aldrich; St. Louis, MO) was made by adding 1 μL of deuterated FA into 1 mL of CD 3 OD. A total of 2 μL of the bioactive fractions were added into 18 μL of acidified methanol above; 15 μL of the mixture were analyzed by direct infusion into a modified 11 Tesla Fourier-transform mass spectrometer (FTMS; Thermo Scientific; Waltham, MA) using a NanoMate robot (Advion; Ithaca, NY) [37]. Full spectra were acquired with resolution set at 100 k.

NMR analysis
Purified bioactive materials were dissolved in 250 μL of CD 3 OD and transferred into a 5-mm Shigemi NMR tube with a glass magnetic plug with susceptibility matched to CD 3 OD on the bottom. All NMR data were collected at 40˚C on an Agilent VNMRS 750 MHz spectrometer equipped with a 5 mm Varian indirect-detection probe with z gradient capability. Collected NMR data included 1 H spectrum, gradient selected correlation spectroscopy (gCOSY), TOCSY, NOESY with a mixing time of 500 ms, 1 H-13 C HSQC spectroscopy, and 1 H-13 C HMBC spectroscopy. The NMR spectra were analyzed using Mnova NMR software (Mestrelab Research, Spain).

Determination of SPF concentration
The proton quantification experiments were performed at 23˚C on an Agilent 750 MHz VNMRS NMR spectrometer equipped with a 5 mm triple-resonance ( 1 H/ 13 C/ 15 N) indirectdetection probe with XYZ PFG gradient capability. The probe was calibrated using the qEstimate tool in the Agilent VnmrJ4.2 software with a known standard. The proton spectrum of the sample was collected with a 90˚pulse angle of 8.5 ms, 16 scans, and 10.4 s delay between scans. The Agilent VnmrJ4.2 software was used to determine the concentration of the sample based on the integration values of proton peaks. A total of 5 well-resolved proton peaks (7.12

Cercarial paralysis assay
To capture the whole field while avoiding excess reflected light in a well, we used the lid of a 96-well plate (Costar; Corning, NY). A total of 40 μL of APW containing approximately 50 cercariae were added to each shallow well on the lid; 10 μL of SPF (dissolved in APW) was then added to reach the final concentration indicated. Using a high-speed camera (Olympus i-SPEED TR) attached to a stereomicroscope (Leica MZ125), we recorded cercariae movement at 20 to 60 fps at 1.25× magnification just prior to addition of test compounds until 3 to 4 min after treatment started. Raw movies were converted to.avi files using i-SPEED Viewer and compressed into JPEG format using ImageJ (addition of compound is considered time 0). We then counted the numbers of free-swimming or paralyzed cercariae every 5 s for 1 min and every 30 s thereafter for 3 min. The number of dead cercariae (those that never swim before and after SPF treatment) were subtracted from data. Experiments were performed in biological triplicate.

Mouse infectivity assay
Swiss Webster mice (female) were purchased from Taconic Biosciences (Rensselaer, NY) and bred by RARC SPF Mouse Breeding Core (University of Wisconsin-Madison, Madison, WI). Mouse infections were performed by exposing mouse tails to S. mansoni cercariae according to standard protocol from BRI [24] with slight modifications. Briefly, we secured mice in rodent restrainers (Thomas Scientific, Cat #551-BSRR, Swedesboro, NJ) and put them vertically on top of a rack with grids. We pipetted 100 μL of each drug at proper concentration into a skinny glass tube (Fisher Scientific, Cat #14-958A, Hampton, NH) inside a 12 × 75 mm holding glass tube (VWR, Cat # 47729-570, Radnor, PA). A total of 300 μL of APW containing approximately 200 cercariae were pipetted into each skinny tube and incubated for 10 min before we inserted the mouse tail. Mouse tails were wiped with APW-moistened Kimwipes, inserted into the skinny tube, and exposed to cercariae for 30 min. The mouse tail was touching the bottom of the test tube unless otherwise specified. Six weeks post infection we euthanized these mice using pentobarbitol and perfused them according to standard protocols [24]. For each drug, we initially used 3 mice for controls (APW only) and 3 mice for each concentration tested except for 25 nM Ht-13-A and Ht-13-A-pr. We then repeated the experiments again with 3 mice for each condition. In addition to that, we included 6 mice for 25 nM Ht-13-A and Ht-13-A-pr.
Adult worms were recovered by hepatic portal vein perfusion, and males and females were unpaired by a brief incubation in 2.5% Tricaine (Sigma-Aldrich; St. Louis, MO) to facilitate counting. We counted total numbers of adult worms under a stereomicroscope (Leica MZ75). Livers from infected mice were fixed in 4% formaldehyde in PBS overnight. The largest liver lobes (left lobes) were submitted to the University of Wisconsin-Madison Histology Core Facility for sectioning and Hematoxylin and Eosin staining. Each left lobe was evenly cut into 4 to 6 pieces and paraffin embedded on a large cassette. One slide (4-6 liver sections) for each liver was used for histological examination, which provided a representative view throughout the whole liver lobe. We took a tiled image of the whole slide using a Zeiss Axio Zoom microscope and used ImageJ to determine the area of each section. Total numbers of eggs in each section were counted and normalized to the area.
In adherence to the Animal Welfare Act and the Public Health Service Policy on Humane Care and Use of Laboratory Animals, all experiments with and care of mice were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Wisconsin-Madison (protocol approval number M005569).

Statistical analysis
GraphPad Prism (version 7) was used for all statistical analyses. One-way ANOVA test followed by Dunnett's multiple comparison test was used. Mean ± SD is shown in all figures.
Supporting information S1 Fig. 1 H NMR spectrum of SPF. Peak areas of the nonoverlapping peaks were integrated and protons (δH 1.34, 3.10, 3.52, 3.56, 3.81, 4.40, 6.86, 6.90, and 7.09) showed integer ratios, supporting the mass spectrometry results that their signals were from the same compound. After adding the integration of overlapping peaks (δH 2.70, 2.72, 2.77, 2.79), a total of 19 protons were discovered, consistent with the best-fitting formula from the mass spectrometry results: (C 16  are joined to a heteroatom. Because proton 18 (δH 2.71) has cross-peaks with both C12 and C15, it is a nitrogen atom that connects methyl group (δC 41.7, δH 2.71 on position 18), CH group (δC 76.7, δH 3.10), and CH 2 group (δC 65.8, δH 2.70 and 3.52). C13 has a chemical shift of 88.2 ppm, suggesting its connection to an oxygen. With HMBC, TOCSY, HSQC, and COSY, the connectivity of the aliphatic portions is resolved. (B) Aromatic region. The connectivity-built aliphatic structure has the formula C 7 H 13 NO, which leaves C 9 H 6 NO after subtracting from the best-fitting formula. HSQC (S3 Fig) showed the existence of a methoxyl group (δC 62.2, δH 3.81). Therefore, the aromatic region was composed of C 8 H 3 N. HMBC data showed that 3 aromatic protons were located in different rings, implying a fused aromatic ring structure with one nitrogen. A substituted indole was the most common structure utilized in organisms with the matching formula. In addition, HMBC showed that protons on the methoxyl group (δH 3.81) and the aromatic proton (δH 6.90) have cross-peaks with carbon (δC 143.1), suggesting they are meta to each other. The other proton (δH 6.86) was vicinal to proton (δH 6.90) because of their coupling seen in the COSY spectrum (S5 Fig). The aromatic singlet proton δH 7.09 showed cross-peaks with 3 aromatic carbons, 2 of those carbons (δC 120.6 and δC 138.1) had cross-peaks with protons (δH 6.86 and δH 6.90), respectively, consistent with an indole configuration. HMBC further confirmed C (δC 110.6) was linked to CH 2 (δH 2.79 and 3.56), and C (δC 143.7) was linked to the CH (δC 88.2, δH 4.40) across an oxygen atom. COSY, correlation spectroscopy; HMBC, heteronuclear multiple-bond correlation; HSQC, heteronuclear single quantum coherence spectroscopy; NMR, nuclear magnetic resonance; SPF, Schistosome Paralysis Factor; TOCSY, total correlation spectrometry. Results of the NOESY experiment support the final structures (Fig 2G and 2H) due to the presence of a NOE signal between H (δH 3.81) and H (δH 1.34), which could only be observed between protons with short spatial distance. For protons on the 3 consecutive chiral centers, H (δH 4.40) had an intense cross-peak with H (δH 2.77), whereas a weak signal was observed between H (δH 4.40) and H (δH 3.10) and no signal was observed between H (δH 2.77) and H (δH 3.10). This suggests that H (δH 4.40) and H (δH 2.77) are close to each other and both are distant from H (δH 3.10), which corresponds to (R, S, S) or (S, R, R) configuration on C 12, 13, 14 (δC 76.7, 88.2 and 37.2). This was further supported by NOESY signals between H (δH 2.79, 3.56) and the 3 H on chiral centers. H (δH 4.40) had a cross-peak with H (δH 2.79) but no cross-peak with H (δH 3.56). However, the opposite was observed for H (δH 3.10), which had a cross-peak with H (δH 3.56) but no cross-peak with H (δH 2.79). NMR, nuclear magnetic resonance; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser effect spectroscopy; SPF, Schistosome Paralysis Factor. (TIF)