Spatial transcriptomics reveals antiparasitic targets associated with essential behaviors in the human parasite Brugia malayi

Lymphatic filariasis (LF) is a chronic debilitating neglected tropical disease (NTD) caused by mosquito-transmitted nematodes that afflicts over 60 million people. Control of LF relies on routine mass drug administration with antiparasitics that clear circulating larval parasites but are ineffective against adults. The development of effective adulticides is hampered by a poor understanding of the processes and tissues driving parasite survival in the host. The adult filariae head region contains essential tissues that control parasite feeding, sensory, secretory, and reproductive behaviors, which express promising molecular substrates for the development of antifilarial drugs, vaccines, and diagnostics. We have adapted spatial transcriptomic approaches to map gene expression patterns across these prioritized but historically intractable head tissues. Spatial and tissue-resolved data reveal distinct biases in the origins of known drug targets and secreted antigens. These data were used to identify potential new drug and vaccine targets, including putative hidden antigens expressed in the alimentary canal, and to spatially associate receptor subunits belonging to druggable families. Spatial transcriptomic approaches provide a powerful resource to aid gene function inference and seed antiparasitic discovery pipelines across helminths of relevance to human and animal health.


Introduction
Lymphatic filariasis (LF) is a chronic and debilitating neglected tropical disease (NTD) recognized as a leading global cause of long-term disability. Over 60 million people are currently infected with LF and~900 million people are at risk of infection across 72 endemic countries [1][2][3]. LF is caused by the filarial nematodes Brugia malayi, Brugia timori, and Wuchereria bancrofti, which reside as adults in the lymphatics producing microfilariae that migrate to the blood and undergo cyclodevelopmental transmission in competent blood-feeding mosquito vectors [4]. Adult stage parasites cause blockage and inflammation of lymphatic vessels that can result in disfiguring and stigmatizing manifestations, including lymphedema (most notably elephantiasis) and hydrocele that afflicts an estimated 36 million individuals [1,2,5,6].
Control of LF relies on routine mass drug administration with anthelmintics, which effectively clear microfilariae but are ineffective against adult stages and are contraindicated in areas co-endemic for other filarial parasites [1,2,7]. Anthelmintic resistance is widespread in veterinary medicine and also represents a threat to filariasis control efforts in both animals and humans [8][9][10][11][12]. To address these challenges and accelerate LF elimination there is a need to generate new antifilarial therapies, particularly drugs effective against adult stage parasites. Current anthelmintics target or dysregulate parasite cell integrity, neuromuscular control, reproductive potential, and the secretion of parasite molecules necessary for the establishment and maintenance of parasitism [13][14][15][16][17][18]. The development of macrofilaricidal (adult-killing) drugs can be hastened by an improved knowledge of tissues that underpin survival in adult parasites.
In adult filarial parasites, vital tissues and interfaces for host-parasite communication are concentrated within the anteriormost region of the body plan. The first millimeter of the B. malayi female head region (~3% of the length) contains cells and tissues that control parasite feeding, sensory, secretory, and reproductive behaviors [19]. Transcriptomic profiling of this region can aid the prioritization of new antifilarial targets, localize the targets of existing drugs, and provide clues to the origins of immunomodulatory molecules released into the host environment. This effort is currently impeded by a lack of scalable transgenesis and in situ localization techniques in this two-host parasite system.
Bulk transcriptomics in filarial parasites has thus far been used to explore changes in gene expression associated with development [20-23] and environmental or drug perturbations [24][25][26][27]. While proteomics has shed light on large and accessible tissues in B. malayi [28], small head-associated structures are massively underrepresented in whole-parasite omics and have yet to be characterized. Here, we adapt spatial transcriptomic and microscopy approaches to profile the head region of B. malayi and resolve gene expression patterns in critical tissues at the host-parasite interface. RNA tomography [29,30] and tissue-specific transcriptomes are leveraged to map the distributions of current drug targets and known antigens, as well as to prioritize putative antiparasitic and vaccine targets. The first application of these complementary methods in a human parasitic nematode provides a template for the localization of gene transcripts and targets of therapeutic and diagnostic value in similarly intractable parasitic nematodes of human and veterinary significance.

The adult filarial head region expresses prominent antigens and known drug targets
Adult stage filariae cause incurable chronic illnesses. To develop new therapies that aid parasite elimination, we must learn more about tissues and structures underlying adult behaviors. Adult female B. malayi are~34.6 mm (31. 8-39.8 mm) in length when reared in Mongolian jirds, but the vast majority of their body plan is composed of mid-body structures including the body wall muscle, the reproductive tract, and intestine [19]. The anterior-most 3% (~1 mm) of the parasite head region contains vital structures including the buccal cavity, amphid neurons, nerve ring, vulva, pharynx, esophageal-intestinal junction, and the excretory-secretory (ES) apparatus. These tissues control essential parasite behaviors and include host-parasite interfaces where drug and antigen interactions likely occur.
To identify head-enriched gene transcripts, individual adult female B. malayi head regions were dissected from the body at the vulva (~0.6 mm from anterior) using ultra-fine probes (Fig 1A). The vulva was chosen as a visible marker to ensure head tissues were captured and isolated from the reproductive tract, which would be contaminated with microfilaria. Lowinput RNA-seq was carried out using paired head and body tissues isolated from three individual parasites. Biological replicates displayed high concordance (Fig 1B), with 70-80% reads from head and body region samples uniquely mapping to the B. malayi genome. Analysis of differentially-expressed genes (DEGs) identified 2,406 head-enriched genes (log 2 (FC) > 1 and p-value < 0.01) with at least 30 total raw reads from the six samples (S1 Table). Transcripts associated with secreted proteins [31,32] are distributed evenly across both head and body region tissues, suggesting mixed origins for what are classically referred to as "ES products" (Fig 1C). In striking contrast, the majority (86%) of prominent filarial antigens with known immunomodulatory capacity, including those that have been pursued as vaccine candidates [33], are head-enriched (Fig 1C). Immunization with recombinant proteins encoded by many genes on this list, including Bm97 and Bma-far-1, confers significant protection against filarial nematode infection in animal challenge studies [34,35].
Antifilarial targets from existing classes of drugs show different distributions (Fig 1D). The putative glutamate-gated chloride channel (GluCl) targets of ivermectin, Bma-avr-14 and Bma-glc-2 [36], show higher relative expression in body tissues consistent with Bma-avr-14 localization to the reproductive tract and developing embryos [37]. Bma-glc-3 and Bma-glc-4 channel subunits are more enriched in the head and may also play a role in macrocyclic lactone responses. Bma-slo-1, a putative target of emodepside, an emerging candidate adulticide for treatment of river blindness [38], is more highly expressed in the body. Conversely, the likely β-tubulin target of albendazole (Bma-btub-1), based on homology to Caenorhabditis elegans ben-1, is head-enriched.

Microscopic investigation of the adult head region and putative excretorysecretory apparatus
intestinal junction (~861-1010 μm) from the anterior have been measured in adult female B. malayi [19], but locating the excretory-secretory (ES) system in adult stage B. malayi and other filaria has been notoriously difficult [39][40][41][42][43]. In microfilariae, the ES apparatus is a hallmark and essential structure consisting of a pore and vesicle leading to a single excretory cell via a cytoplasmic bridge [44]. Ivermectin is thought to disrupt microfilarial ES protein and exosome release through binding to ion channels in the vicinity of the ES vesicle [14,16,17]; however, these structures become inconspicuous through development [39,40,42,45,46]. To help pinpoint the ES in adult female B. malayi, the relative organization of head structures across Clade III [47,48] nematodes was collected from available literature (Fig 2A). Among Clade III parasites, the ES pore and/or cell are located posterior to the nerve ring in 23/24 species, anterior to reproductive openings (24/24 species), and anterior to the esophageal-intestinal junction (23/24 species) in at least one life stage surveyed. The conservation of structural organization across developmental and evolutionary time indicates the presence of ES structures between the nerve ring and vulva in adult female B. malayi.
To identify the ES pore in adults we optimized live 4D light sheet microscopy as well as multiple electron microscopy methods. Critical point drying scanning electron microscopy (SEM) of adult Brugia and the closely-related but much larger filarial parasite Dirofilaria immitis allowed clear visualization of the vulva, but not the ES pore (S1 Fig). This may be

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode due to the small size and angle of the pore opening [49]. Light sheet imaging was adapted for live adult males partially paralyzed with 1 mM levamisole to restrict gross muscle movement, and adults were monitored for up to 1 hr at 10 s intervals. Males were chosen to avoid confusion with the confounding activity of the vulva, which is proximal in females. Nuclei stained head regions revealed instances of pulsing during which stain condensed into a large cell with a pore-like tubular structure that was then cleared from the worm~430 μm from the anterior (Fig 2B and S1 Video). This location is consistent with the ES pore location (397-537 μm) in the fur seal parasite Acanthocheilonema odendhali [50], the only filarial nematode where the adult stage ES pore has been morphometrically characterized. To our knowledge, these pulses represent the first evidence of dynamic ES pore opening events in a mammalian parasitic nematode.
To obtain a finer description of head structures and potential ES channels, we utilized highpressure freeze fixation with serial block face-SEM (SBF-SEM) to obtain approximately~1000 serial sections (~70 nm/section) from the anterior of an adult female (Fig 2C, see also data deposited on https://doi.org/10.6084/m9.figshare.16441689.v1). The ventral nerve cord and pharynx were present throughout and we observed 30 pharyngeal, 21 body wall muscle (~5 per muscle quadrant), 5 ventral nerve cord, and 83 lateral cord (~40 per cord) nuclei. Similar to C. elegans, the lateral cords appear to be partially composed of epidermal syncytia with multiple closely apposed nuclei (Fig 2C) that were evident along the anterior-posterior axis in SBF-SEM sections 1-37, 52-92, 189-250, 431-449, 516-543, 660-683, 737-800. We did not observe any nuclei within the dorsal cord itself. The C. elegans excretory canal [51] is visible in EM sections immediately ventral of the lateral cords, while in Onchocerca volvulus a glomerulus-like excretory-structure [52] is suggested to be embedded within the lateral cords. Neither canal-type was observed in our SBF-SEM data. Their absence is possibly due to individual variation in the position of ES structures or the excretory system may be greatly reduced in size, as proposed previously [40,52].
Within lateral cords we observed membrane bound processes along the lateral and basal edges. These processes were described previously as axons or infolded membranes [43,53,54]. In some regions processes are also embedded within the lateral cord, while others appear to bisect the lateral cord. Similar processes embedded in the lateral cord are not seen in C. elegans [55]. To better define the lateral cord in this region we turned to transmission electron microscopy (TEM) (Fig 2D). As previously suggested, membrane processes embedded within the lateral cord appear to be neuronal, as evidenced by numerous microtubules. However the processes located on the basal boundary of the lateral cord lacked microtubules. The absence of consistent microtubules argues against a solely commissure identity. Another possibility is that these structures towards the interior of the worm comprise a modified excretory system. Their position adjacent to the pseudocoelom would be consistent with an excretory system; however, additional serial data are needed to identify the nature of these structures.
TEM also demonstrated the presence of likely homologs to the C. elegans seam cells along the median ridge surrounded by putative epidermal syncytia (Fig 2D). These are readily identifiable by their position and the presence of adherens junctions connecting the seam to the syncytial epidermis. In C. elegans and other nematodes, the seam cells have stem cell-like properties and act to contribute nuclei to the growing epidermal syncytia [56,57]. As previously shown, the lateral cords were also enriched in Wolbachia endosymbionts [54]. on the median ridge of the lateral cord and the presence of adherens junctions (arrows) connecting to surrounding epidermis. Scale bar = 2 μm. (d") Closeup of membranous process enriched in microtubules and surrounded by epidermis. Scale bar = 400 nm. https://doi.org/10.1371/journal.ppat.1010399.g002

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode

Spatial transcriptomics maps antiparasitic targets associated with essential tissues
To deconvolute gene expression patterns across the adult female B. malayi head-region we adapted and optimized RNA tomography from model organisms [29,30]. Individual adult females were oriented and cryo-embedded for collection of 20 μm sections along the anteriorposterior axis. Cryosection imaging was used to validate the tissue collection protocol (Fig 3A) and generate estimates of nuclei density at 20 μm resolution across the targeted region ( Fig  3B). Each cryosection contains a mean of 10.15 (±1.46) nucleated cells anterior to the vulva, with decreased cell densities at the anterior tip and prior to the appearance of the ovarian tract. These observations correspond to live light sheet imaging data, with approximately 9.70 and 9.24 nuclei per 20 μm in females and males, respectively (Fig 3C).
Single-worm RNA tomography was performed via sequential capture and 96-well plate based processing of individual 20 μm sections for low-input RNA-seq (Fig 4A). Read mapping rates are negligible (< 1%) through the first 11 sections, reflecting a conservative capture strategy to avoid missing the anterior tip of the head, and rise to 70% (± 14%) through the remaining sections. Hierarchical clustering of three RNA tomography replicates reveals unique gene expression signatures across sections (Fig 4B and 4C). Robust genes were defined as those associated with >20 counts across all sections for a given replicate and >10 counts in a single section. Robust sections were defined as those expressing at least 100 genes with >10 counts.

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode High-quality cryosections associated with high rates of uniquely mapped reads fall below the blue line (cryosection 12). Z-scores reflect scale-normalized TPM counts. (C) Spatial gene expression heatmaps for additional replicates (R2 and R3). (D) Cumulative protein-coding genes identified along the anterior-posterior axis for each replicate. Dashed line represents the estimated total protein-coding gene count for B. malayi. Genes were included in the count if they were found in at least one slice with > 10 raw reads. (E) Overlap of robustly-expressed genes detected across replicates. (F) Clustering of spatial expression patterns for robustly-expressed genes detected across replicates. Data are shown for the highest quality RNA tomography run (x-axis: cryosections 12-48). Hierarchical clustering was carried out using a distance matrix derived from scale-normalized TPM counts. Cluster IDs (S2 Table) are shown for both localized (green) and diffuse (orange) spatial expression patterns (blue line represents mean expression). Localized patterns were defined as those with a single major peak restricted to a single cryosection. The three highest expressed genes are provided as markers for peaks localized to a single cryosection. https://doi.org/10.1371/journal.ppat.1010399.g004

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode 8,900 genes are detected as robustly expressed across the head of the first (highest-quality) replicate (Fig 4D). Additional replicates exhibited lower gene coverage, but shared 97% and 94% of their detected genes with the first run (Fig 4E). 5,810 genes were detected as robustly expressed across all replicates, including 2,375 of 2,406 (98.7%) transcripts previously identified as head-enriched. RNA tomography captured approximately 48% of B. malayi proteincoding genes, likely reflecting the great diversity of tissue and cell types contained within the relatively small head region.
Genes were grouped by spatial expression pattern, displaying either localized or diffuse expression patterns down the anterior-posterior axis. The former likely represent gene transcripts and markers restricted to distinct neurons, while the latter reflect recurring cell types such as the epidermis, body wall muscle, or pharynx (Fig 4F and S2 Table). Prominent secreted antigens (e.g., Bma-mif-1, Bma-tpi-1, Bma-cpi-2), including proteins associated with exosomes (e.g., Bma-lec-1 and Bma-enol-1) [17], do not fall into a specific cluster, reinforcing the heterogeneous nature of their transcriptional origins even if ultimately released from the same orifice. Neighboring cryosections were coalesced to map the most abundantly expressed genes with respect to head structures. The most highly expressed pre-vulval genes are the neuropeptide-like protein Bma-nlp-77 and the collagen Bma-col-72, replaced by the immunogens Bma-val-1 [58] and Bm97 [35] where the vulva is expected to appear (Fig 5A). A fraction of post-vulval transcripts are likely associated with progeny in the reproductive tract.
We next examined the spatial distributions of druggable receptor and ion channel families, detecting 10 (of 11) transient receptor potential (TRP) channel subunits, 43 (of 52) cys-loop ligand-gated ion channel (LGIC) subunits, and 10 (of 11) aminergic G protein-coupled receptors (GPCRs) across the head region (TPM > 10 in at least one cryosection). A subset of these receptors are restricted to or enriched in the vicinity of the nerve ring, ES pore, or vulva ( Fig  5B), suggesting an outsized role in neuromuscular control of movement, secretion, or fecundity.
LGICs represent the most successfully targeted class of anthelmintics, mediating the antiparasitic effects of nicotinic receptor agonists and macrocyclic lactones. These pentameric channels can be pharmacologically characterized in heterologous cells [59] but it is unknown whether heteromeric channels functionally constituted in surrogate systems reflect endogenous channel subunit interactions. To guide heterologous studies, we used spatial correlations among channel subunits for major LGIC subfamilies (S2 Fig) to predict subunits that are more likely to be found in the same cells and form functional channels (Fig 5C).

Discovery of candidate hidden antigens in the pharynx and intestine
The nematode pharynx and intestine are established target sites for existing [59][60][61] and emerging anthelmintics [62] and also act as critical host-parasite interfaces with potential sources of 'hidden' antigens for vaccine development [28, 63,64]. These antigens may evade host immune recognition but remain accessible to vaccine-induced antibodies in the course of parasite feeding [64]-the rationale behind the protective immunity offered by the commercial Haemonchus contortus vaccine in ruminants [65]. While RNA tomography provides an anterior-posterior map of gene expression across the head region, these data alone cannot be used to cleanly infer intestinal or pharyngeal transcriptomes. The alimentary tract runs the length of the body and is one of a few tissues which can be effectively isolated by microdissection. However, the pharynx cannot be easily isolated and likely contains a unique expression profile in comparison to the intestine. To profile the pharynx, we developed a laser capture microdissection (LCM) approach to capture pharyngeal cells from cryosectioned head tissue.
To capture whole intestines, live adult female B. malayi individuals were microdissected by removing the head and tail and separating the intestine from other tissues with minuten pins

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode (Fig 6A). RNA sequencing of intestinal and non-intestinal tissues revealed 1,077 intestineenriched genes (log 2 (FC) > 1 and p-value < 0.01) with 489 genes predicted to contain at least one transmembrane domain. Among these, 64 putative membrane proteins were further prioritized as candidate hidden antigens based on high intestinal expression (mean TPM > 100) and relatively low abundance in non-intestinal tissues (intestinal:non-intestinal TPM ratio > 10) (Fig 6B and S3 Table). These data greatly expand on the Brugia intestinal proteome [28] and provide new leads that are more likely to be tissue restricted. We identify cathepsin-like protease Bma-cpl-1 as an intestinally-enriched target, along with membrane targets that include a GABA receptor subunit (Bma-gab-1), glutamate transporter (Bma-gtl-1), neuropeptide GPCR (Bma-
To determine whether targets in the intestine were conserved in the upper alimentary tract we performed LCM of pharyngeal tissue. Pharyngeal and adjacent non-pharyngeal tissue were isolated from one another by LCM using UV cutting and IR capture lasers on 10 μm thick adult female head sections generated by the RNA tomography cryosectioning technique (Fig 6A). Each collected sample, estimated to contain fewer than five cells, was subject to low-input RNAseq. Pharyngeal tissue samples cluster as expected, while non-pharyngeal samples are divergent, reflecting collections of disparate cell types from different positions in the head (S3 Fig). Focusing on transmembrane proteins, we observe little overlap between the most highly expressed genes in the pharynx, including those more restricted to the pharynx (pharyngeal:nonpharyngeal TPM ratio > 10), and prioritized intestinal targets (Fig 6D and S4 Table). This

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode suggests that the upper and lower alimentary canal are sources of unique targets and candidate hidden antigens. Bma-hpo-8 and Bma-pcp-1, a membrane-bound peptidase, represent exceptions enriched across the alimentary canal. Both intestinal and pharyngeal hidden antigen candidates are composed mostly of genes of unknown function and many are orthologous to extra-intestinal C. elegans genes. This highlights the need for care in ascribing functions and spatially mapping tissue-specific markers from this clade V model nematode to distantly-related clade III parasites.

Discussion
Spatially resolved gene expression patterns and tissue-specific transcriptomes can aid our functional understanding of genes [67], especially in non-model organisms where transgenesis and functional genomics are not routine. To this end, we have generated the first genomewide expression maps at fine scale in a multicellular parasite, focusing on the B. malayi adult head region. This tightly organized space encompasses tissues and structures responsible for vital sensory, secretory, reproductive, and feeding processes required for parasite survival and the maintenance of infection in the human host. We adapted low-input tissue capture and RNA tomography, combined with light-sheet and electron microscopy, to survey anterior-posterior expression patterns and map these data to tissues of interest.
Openings in the B. malayi head act as druggable host-parasite interfaces and as potential conduits for prominent secretory antigens [31,68] and vaccine candidates [28,33,69]. Secretomes of adult stage B. malayi are well defined [13,31] but the anatomical origins of these products are unknown, owing in part to difficulty identifying the adult ES-pore [39,40,42,43,45,46]. We show that the great majority of prominent antigens are head-enriched, but do not fall into a specific spatial pattern within the head region. Complementary light sheet and electron microscopy efforts provide the first direct evidence of ES pore activity in a mammalian parasitic nematode and suggest a potentially contracted or modified ES system in the adult stage [40,52], which requires further investigation.
Spatial and tissue-specific transcriptomics data were leveraged to map the distributions of current anthelmintic targets and to prioritize new drug and vaccine targets, including candidate membrane-anchored 'hidden' antigens that are highly-expressed and more likely to be restricted to the upper or lower alimentary canal. Transcripts encoding for proteins that belong to traditionally druggable receptor families were associated with the estimated locations of the nerve ring, ES pore, and vulva. These receptors may serve as targets for the dysregulation or inhibition of parasite neuromuscular control, host-parasite communication, and fecundity.
While spatial transcriptomics provide positional information in the context of a whole organism or region, they do not capture individual cells and replicates are challenging to align. Differences in replicates can be attributed to both biological and technical sources of variation, but we speculate that biological variation is the primary driver of these differences. Adult stage worms exhibit significant anatomical size variation and it is expected that different groups of cells are collected together in 20 μm sectioning windows across replicates. To complement RNA tomography, single-cell approaches in parasitic nematodes, such as those applied in C. elegans [70], can provide more granular information about cell and tissue-specific expression patterns. However, unlike C. elegans, there are no validated markers to map cell lineages in B. malayi. Transgenic approaches are developing [71], but cannot conceivably be scaled given the challenges of the two-host life cycle. Ultimately, the integration of spatial and single-cell data from B. malayi will provide a pathway to marry transcription to specific cells within defined tissues of interest.
Taken together, these findings highlight the utility of spatial transcriptomic techniques when applied in parasitic nematodes and show that the resulting data can be used to define region and tissue specific gene expression patterns in small and densely packed parasite tissues.

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode

Parasite culture
Brugia malayi adults (NIH-FR3) extracted from the Meriones unguiculatus infection system were maintained in daily changes of RPMI 1640 with L-glutamine (Sigma-Aldrich) supplemented with FBS (10% v/v, Fisher Scientific) and penicillin-streptomycin (100 U/mL, Gibco) at 37˚C with 5% CO 2 unless otherwise specified. Individual adults were separated by sex into 3-4 mL of culture media. For RNA-seq analyses, individual worms were acclimated in culture for 18-24 hrs prior to fixation, preservation, or RNA extraction unless otherwise stated.

Parasite tissue dissections
A modification of the Morris et. al.
[28] method was employed where individual adult female B. malayi were washed 3x in nuclease-free PBS then dissected using Eliminase (VWR) cleaned 0.15 mm minuten pin dissecting probes (Bioquip) in PBS in a petri dish under a ZEISS Stemi 508 with Sony Exmor CMOS IMX178 camera. For head vs body RNA-seq, the head was severed by crossing two probes adjacent to the vulva. For the intestine vs carcass RNA-seq, the body was held in place using one probe and another was used to gently nick and pierce the cuticle at the midbody, releasing internal organs. The head and tail were then severed to free intestinal tract ends. Eliminase washed No. 5 forceps were used to pull the intestine away from the body. Individual tissues were transferred to 2 mL Safe-Lock tubes (Eppendorf) containing 300 μL TRIzol LS (Invitrogen) and 100 μL nuclease-free water, vortexed, flash frozen in liquid nitrogen, and stored at -80˚C. For RNA extraction, samples were thawed and homogenized with a single 3 mm Eliminase washed stainless steel ball bearing in for 3 min at 30 Hz (TissueLyser II, Qiagen) then purified by the Direct-zol RNA microprep kit (Zymo).

RNA tomography: Tissue preparation and RNA extraction
Individual adult female B. malayi were washed thrice in RPMI 1640 with L-glutamine, soaked in RPMI 1640 with 0.005% methylene blue for 2 min, then washed once more with RPMI 1640. Stained worms were picked using Eliminase cleaned 0.15 mm minuten pin dissecting probes directly into clear TissueTek O.C.T. (Electron Microscopy Sciences) in a Stainless Steel Base Mold (Simport Scientific). The mold was positioned immediately prior to staining on a bed of dry ice under a ZEISS Stemi 508 with Sony Exmor CMOS IMX178 camera. Once in O. C.T., the body was straightened so that the head region was positioned parallel to the long face of the mold using the probe. A~0.5 μL methylene blue (0.1% in water) dye dot was then placed roughly 1 mm above the anterior-most portion of the head to act as a location marker. During freezing the position of the worm was imaged in relation to the edge of the mold and the dye marker to calculate distance between the block edge and the dye dot as well as the dye dot to the sample. Frozen O.C.T. blocks were covered in parafilm, indexed, and stored at -80˚C until sectioning. Cryosections (48 x 20 μm) along the anterior-posterior axis were taken on a Leica

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode

Laser capture microdissection (LCM) of the adult female pharynx
Sections (10 μm) were collected using the RNA tomography sectioning protocol and placed directly onto UV irradiated (254 nm, Stratalinker) PEN Membrane Glass Slides (Applied Biosystems). Slides were rinsed once in nuclease-free water for 1 min to remove O.C.T, followed by a 1 mL wash series in 70%, 90%, 95%, and 100% ethanol, then air dried for 10 min and stored at -80˚C in a 50 mL conical tube. Slides were equilibrated to room temperature before being loaded onto an ArcturusXT LCM Instrument. Sections were inspected for the presence of a pharynx, and captured under CapSure HS LCM Caps (Applied Biosystems) by UV laser (settings include UV current: 15, cutting speed: 300, pulse frequency: 500, section thickness: 10, cut: 10, tab length: 1) and captured by IR (settings include IR spots: 3, spacing: 60, diameter: 50-75, power: 99, duration: 49). Cuts were made to separate the pharynx from the rest of the tissue, leaving a tab of tissue that connects to the rest of the PEN membrane. Each cut section was collected on one CapsSure cap with the remaining tissue collected on another cap in GeneAmp Thin-Walled Reaction Tubes (Applied Biosystems) and transferred to dry ice for immediate processing after collection. Three replicates were performed with 2-4 sections per replicate per group (pharynx vs non-pharynx).

RNA-seq library preparation and sequencing
For NEBNext library preps, DNA quantity was checked by Qubit (dsDNA HS Assay Kit, Invitrogen) and SPRIselect beads (Beckman Coulter) were used for DNA purification steps. For cDNA amplification and PCR enrichment of the adapter ligated DNA, cycle numbers were optimized for each sample unless stated otherwise.
Head vs body. The Clontech SMARTSeq v4 Ultra-Low Input RNA kit (Takara) was used with 1.7 ng of input RNA from each sample, as determined by Agilent RNA 6000 Pico Kit on a 2100 Bioanalyzer (Agilent). Full length cDNA was quantified by 2100 Bioanalyzer. 150 pg of amplified cDNA was tagmented and index-amplified using Nextera XT adapters (Illumina). DNA quantity was assessed by Qubit (dsDNA HS Assay Kit) and quality by 2100 Bioanalyzer. Libraries were balanced by Illumina MiSeq Nano for a single lane of 1x100 bp sequencing on the Illumina HiSeq 2500.
RNA tomography. The first library was prepared as described for head vs body RNA-seq. For additional replicates, 8 μL of RNA was added per section to step 2.1 through to 2.11.11 of the NEBNext Single Cell/Low Input RNA Library Prep Kit for Illumina (NEB, version 3.0, #E6420L) using NEBNext Multiplex Oligos for Illumina (Index Primers Set 1, NEB, #E6440G) and sequenced on an Illumina Novaseq 6000 (2x150 bp, S1 flow cell).
Intestine dissection and pharynx LCM. Libraries were prepared as described above with the NEBNext Single Cell/Low Input RNA Library Prep Kit for Illumina (NEB, version 3.0, #E6420L). For intestine and carcass tissues, 8 μl of purified RNA was added per sample to step 2.3 through to 2.11.11. For pharynx and non-pharynx RNA, tissues on CapSure caps were transferred by Eliminase cleaned No. 5 forceps directly to 8 μl of 1x NEBNext Cell LysisBuffer, frozen at -80˚C overnight for one cycle of freeze-cracking to release RNA. Lysate was submitted to step 1.3 through to step 1.12.11. For pharynx and non-pharynx RNA, 20 cycles of cDNA amplification were performed at step 1.5 and 12 cycles of PCR enrichment of the adapter ligated DNA were performed at step 1.11. Libraries were sequenced on an Illumina Novaseq 6000 (2x150 bp, 4 million reads per sample).

Epifluorecent imaging of representative B. malayi cryosections
Sections (20 μm) were collected using the RNA tomography sectioning protocol, placed sequentially in rows on charged slides (Thermo Scientific) and allowed to dry at room

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode temperature for 5 min. Slides were rehydrated in PBS for 30 s followed by marking of sections with diamond pen, rinsed in molecular grade water and air dried again. Sections were then fixed in 3.4% formaldehyde at room temperature for 15 min, washed twice with PBS and stored in 70% ethanol. To estimate cell numbers, sections DAPI (Invitrogen) to mark DNA and phalloidin (Invitrogen) to mark actin. Slides were stained with 20 mL of 70% ethanol containing 1 μg DAPI and 5 μL AlexaFluor 488 Phalloidin for 8 min at room temperature, washed in 70% ethanol, then rehydrated and mounted in PBS prior to imaging on a Zeiss Axio Scope A1. 3D rendering of color-merged sections was performed in Fiji [72] using TrakEM2 [73] to orient all sections per individual as a stack, which was then compiled in 3D viewer [74] (voxel depth 124 / 20 μm section, resampling rate = 1).

Serial block face-SEM (SBF-SEM) and TEM
Adult female B. malayi were prefixed in 2% paraformaldehyde and cut with a scalpel posterior of the vulva. The anterior portion was immediately placed into a 3 mm specimen carrier with 20% bovine serum albumin (Sigma-Aldrich) and cryofixed in a high pressure freezer (Alba HPM 010). Cryofixed specimens were transferred into a RMC FS-8500 freeze substitution system for substitution in2% OsO 4 , 0.1% uranyl acetate and 2% H 2 O followed by en bloc Osmium-Thiocarbohydrazide-Osmium staining [75,76] and embedding in Durcupan ACM (Electron Microscopy Sciences). Serial block-face imaging was conducted in a Zeiss Sigma 3View system with variable pressure with 30 nm x 70 nm resolution at 5.5 kV. Specimens for TEM were processed and freeze substituted as described above followed by embedding in Poly/Bed 812 (Polysciences) and sectioned on a RMC PowerTome ultramicrotome at 70 nm thickness. Sections were stained with lead citrate and uranyl acetate and imaged on a Phillips CM200 TEM.

Light sheet microscopy
Adult male or female B. malayi were individually incubated at 37˚C in RPMI 1640 with L-glutamine with 125 nM DRAQ5 (Biolegend) for 24-48 hrs, washed once in RPMI 1640 and once in ddH 2 O, then immediately transferred to 37˚C 1-1.2% low melt-temp agarose (Sigma) with 1 mM levamisole hydrochloride (�99%, TCI America). Individuals were mounted in FEP tubing with males in 0.8 mm inner diameter (BOLA) and females in 1.6 mm inner diameter tubes (BOLA) according to published protocols [77]. Individuals were imaged on a customized multi-view light sheet microscope similar to a previously published system [78]. The light sheet was created using a cylindrical lens and projected into the sample via an illumination objective (Olympus #UMPLFLN10XW, 10x/0.3). The fluorescence signal was collected with another objective (Olympus #UMPLFLN20XW, 20x/0.5) perpendicular to the illumination objective. A fiber-coupled laser engine (Toptica MLE) was used as the laser source delivering excitation light at 640 nm. Images were processed into maximum intensity projections (MIPs) and stacked into time-series using Fiji [72]. Nuclei counts and location (X,Y coordinates in relation to the centrepoint of the head tip) were collected manually from individual MIPs using the multi-point tool in Fiji [72].

Bioinformatic analyses
Short-read RNA sequencing data was trimmed using fastp [79] and aligned to the B. malayi reference genome (Worm Base Para Site [80], release 15) using STAR [81]. The RNA-seq pipeline was implemented using Next flow [82] and is publicly available (https://github.com/ zamanianlab/Core_RNAseq-nf). All downstream expression analyses were carried out using a mixture of custom R, bash, and Python scripts, including hierarchical clustering and visualization. Identification of robustly-expressed genes and cryosection quality control were carried out using raw gene counts, while hierarchical clustering and primary heatmap analyses were performed with scale-normalized TPM values. Differential expression analyses were carried out using DESeq2 [83]. Genome-wide transmembrane prediction was performed with HMMTOP v2.1 [84]. B. malayi cys-loop ligand gated ion channels (LGICs) were identified using a reciprocal blastp [85] and profile HMM [86] approach using a database of known C. elegans LGICs. Ion channel subunits were aligned with MAFFT [87] and trimmed with trimAl [88] such that columns with greater than 30% gaps were removed, and sequences that did not have at least 70% of residues that aligned to columns supported by 70% of the sequences were removed. The trimmed, filtered alignment was subjected to maximum-likelihood phylogenetic inference with IQ-TREE 2 [89] and ModelFinder [90] with ultrafast bootstrapping [91], using the VT substitution matrix [92] with empirical base frequencies and a free-rate substitution model [93] with 10 categories. Bootstrap values from 1,000 replicates were drawn as nodal support onto the maximum-likelihood tree.

PLOS PATHOGENS
Spatial transcriptomics identifies antiparasitic targets in a human parasitic nematode