Tissue Specific Profiling of Females of Schistosoma japonicum by Integrated Laser Microdissection Microscopy and Microarray Analysis

Background The functions of many schistosome gene products remain to be characterized. A major step towards elucidating function of these genes would be in defining their sites of expression. This goal is rendered difficult to achieve by the generally small size of the parasites and the lack of a body cavity, which precludes analysis of transcriptional profiles of the tissues in isolation. Methodology/Principal Findings Here, we describe a combined laser microdissection microscopy (LMM) and microarray analysis approach to expedite tissue specific profiling and gene atlasing for tissues of adult female Schistosoma japonicum. This approach helps to solve the gene characterization “bottle-neck” brought about by acoelomy and the size of these parasites. Complementary RNA obtained after isolation from gastrodermis (parasite gut mucosa), vitelline glands and ovary by LMM were subjected to microarray analyses, resulting in identification of 147 genes upregulated in the gastrodermis, 4,149 genes in the ovary and 2,553 in the vitellaria. Conclusions This work will help to shed light on the molecular pathobiology of this debilitating human parasite and aid in the discovery of new targets for the development of anti-schistosome vaccines and drugs.


Introduction
Members of the genus Schistosoma are parasitic blood flukes responsible for the serious but neglected human disease of schistosomiasis [1,2]. In common with other platyhelminths, schistosomes exhibit acoelomy, the body plan characteristic of basal bilaterians whereby tissues are bound together by cells and matrices of the parenchyma in the absence of a body cavity. This body organization, together with the generally small size of adults and developing stages, has been a major hindrance for functional analyses of individual schistosome tissues and cells, because it has been impossible to isolate them. These problems are exacerbated by poor knowledge and limited annotations of many schistosome genes and the absence of basic knowledge of where, and when, in development the molecules are expressed. Localization methods incorporating immunocytochemistry and in situ hybridization have been at the vanguard of functional studies of schistosome proteins [3], but the prospect of obtaining robust, informative localization data of multiple genes expressed throughout the complex schistosome life cycle remains a daunting challenge.
Concerted international efforts have been directed at defining functional relevance of the predicted 14-16,000 schistosome genes to identify potential targets for drug and vaccine therapies [4,5].
Here, we report on tissue-specific gene expression analysis of adult female Schistosoma japonicum, as a means to expedite functional characterization of schistosome gene products. Our approach incorporates methods of laser microdissection microscopy (LMM) to generate tissue-specific transcriptional extracts for subsequent microarray analysis. This work follows hypotheses [19,20] that LMM would prove an excellent means to expedite transcriptional typing of many schistosome tissues despite the acoelomate body plan of these parasites. sis hupensis snails, collected from Anhui Province, China, were provided by the National Institute of Parasitic Diseases-CDC, Shanghai. Adult worm pairs were perfused 6 weeks post-challenge from infected ARC Swiss mice. Two batches of approximately 25 live female parasites were flat embedded in Tissue-Tek Optimal Cutting Temperature compound (OCT) (ProSciTech, Australia) and snap-frozen on dry ice. The sample blocks were stored at 280uC, prior to sectioning with sterile blades in a cryostat. Sections were cut at 7 mm and mounted onto a sterilized polyethylene-naphthalene membrane on a microscope slide (P.A.L.M. Microlaser Technologies, Germany). The slides were then stored at 280uC.
For transmission electron microscopy, female parasites were fixed in 3% glutaraldehyde in 0.1 M phosphate buffer at pH 7.4 for 2 h, post-fixed in potassium ferricyanide-reduced osmium tetroxide, followed by 5% aqueous uranyl acetate, dehydrated in acetone and embedded in Epon resin. Ultrathin sections were viewed on a JEM 1011transmission electron microscope operated at 80 kV.

Laser microdissection microscopy (LMM)
Thawed cryo-sections were fixed immediately in 100% methanol for 30 seconds, stained with 1% Toluidine blue (CHROMA, Germany) for 10 seconds, washed in diethylpyrocarbonate (DEPC)-treated water, 2610 seconds, and allowed to dry for 20-30 min before microdissection. A PALM microbeam laser catapult microscope (P.A.L.M. Microlaser Technologies) was used to microdissect the gastrodermis from posterior regions of female worms, and the ovary and vitelline tissues from the stained frozen sections. An area of approximately 4 million squared mm (approximately 20610 6 mm 3 of tissue) was collected separately from each of the tissues onto 500 ml opaque adhesive caps (P.A.L.M. Microlaser Technologies). The areas amount to the collection of many 1000's of microdissected elements for each tissue. For control tissue, we used 12 S. japonicum females that were snap-frozen in OCT and sectioned by cryostat. Control sections were collected onto 6 sterile glass slides. Entire sections were then scraped by sterile scalpel blades from the slides into RNA extraction buffer (below) for analysis. The control samples therefore represent the transcriptional repertoire of entire females.

Total RNA isolation and hybridisation
Total RNA was isolated from the control and LMM samples using RNAqueous-Micro kits (Ambion) kit using the manufacturer's instructions and quantified using a Nano-Drop ND-1000 spectrophotometer (Thermo Scientific, USA). The quality of total RNA was assessed using a Bioanalyser RNA Pico Lab Chip (Agilent) prior to storage at 280uC.

Microarray hybridisation and feature extraction
Full details of the design and construction of the schistosome microarray used have been reported [11]. In brief, the array was constructed from information based on the transcriptomes of adult S. japonicum and S. mansoni. The microarray consists of 19,222 target contiguous sequences (contigs) printed twice from two independent probe designs, and includes 12,166 probes derived from S. mansoni, and 7,056 probes derived from S. japonicum. An overview of the design and composition of the microarray is present in Table S1.
A 300 ng aliquot of total RNA from each sample was converted into complementary RNA was synthesized and labeled with the fluorophore Cyanine 3-CTP (CY3c) and hybridized according to the manufacturer's instructions (Agilent Technologies -One-Color Microarray-Based Gene Expression Analysis). Microarray hybridisations were performed in duplicate for all samples. The yield, concentration, amplification efficiency and abundance of CY3c were measured at A260 and A550 by spectrophotometry.

Data analysis
Hybridized slides were scanned using an Agilent Microarray Scanner (B version) as tiff files and processed with the Feature Extraction 9.5.3.1 Image Analysis program (Agilent) to produce standardised data for statistical analysis. All microarray slides were assessed for background evenness by viewing the tiff image by Feature Extraction. Feature extracted data was analysed using GENESPRING (version 7.3.1; Agilent Technologies/Silicon Genetics, Redwood City, CA).
Microarray data were normalised using a normalisation scenario for ''Agilent FE one-color'' which including ''Data Transformation: Set measurements less than 5.0 to 5.0'', ''Per Chip: Normalize to 50th percentile'' and ''Per Gene: Normalize to median''. Data sets were further analysed using published procedures based on one-colour experiments [21]. The gProces-sedSignal values determined in GENESPRING using Agilent's Feature Extraction software including aspects of signal/noise ratio, spot morphology and homogeneity. Thus, gProcessedSignal represents signal after localised background subtraction and includes corrections for surface trends. Features were deemed Absent when the processed signal intensity was less than two fold the value of the processed signal error value. Features were deemed Marginal when the measured intensity was at a saturated value or if there was a substantial amount of variation in the signal intensity within the pixels of a particular feature. Features that were not Absent or Marginal were deemed Present. Data points were included only if Present or Present, Absent and probes or contigs retained if all data points were Present or Present, Absent.
Microarray data have been submitted to the Gene Expression Omnibus public database, under accession numbers GPL7160 and GSE12706.

Author Summary
Schistosomes are parasitic worms responsible for important human diseases in tropical and developing nations. There is urgent need to develop new drugs and vaccines to augment current treatments for this disease. In recent years, concerted efforts by many laboratories have led to extensive genetic sequencing of the parasites, and the publication of genome sequence for two agents of schistosomiasis appears imminent. This genetic information has revealed many molecules expressed by the schistosome parasites for which no functional information is available. This lack of information extends to ignorance of where in the complex multicellular schistosome parasites the genes are expressed. We integrated two molecular and cellular techniques to address these knowledge gaps. We used laser microdissection microscopy to dissect small but highly important tissues involved in nutrition and reproduction from sections of female Schistosoma japonicum. From these dissected tissues we then used a broad molecular biology method to identify the multiple genes active in these tissues. Our approach has allowed us to formulate the basis of a ''gene atlas'' for schistosome parasites, defining the expression repertoire of specific tissues. The better understanding of the roles of tissues in parasite biology, especially in development, reproduction and interactions with its human hosts, should promote future investigations into pathogenesis and control of these significant parasites.

Gene ontology analysis
Batch BlastX (6 frame translation protein homology) was performed at http://www.blast2go.de on all contigs. This presented a further overview of the gene ontologies that are modulated between tissue types in adult female S. japonicum ( Figure  S1 and Table S2). This information was used to supplement previously published GOs based on nucleotide sequence [11]. To gain a more complete overview of the GO categories that are modulated during the S. japonicum lifecycle we used the software ErmineJ to produce extended list of GOs associated with each of the microdissected tissue types [22].

Real time PCR validation of microarray data
A total of 9 gene sequences indentified as differentially expressed among the three S. japonicum tissues and whole worm control tissue were chosen for validation of microarray data using real time PCR as described [12]. The template for real time PCR was that obtained by microdissection. Forward and reverse primers (Sigma-Aldrich, Australia) were designed from the 10 contigs (Table S3). All total RNA samples were DNase treated (Promega, Australia) prior to synthesis of cDNA using a QuantiTect Whole Transcriptome Kit (QIAGEN, Australia). All cDNA samples were diluted to a concentration of 5 ng/ml. Real time PCR was performed in a Gene Disc 100 ring (Corbett Research, Australia). A sequence from the NADH-ubiquinone reductase gene of S. japonicum was used for normalisation of data from all experiments. Each experiment was performed in duplicate, and the confidence threshold (CT) of the second set was normalised to the first set before evaluation. This was done by importing the standard curve of the first set to that of the second using Rotor Gene 6 software [12]. Microarray and real time PCR datasets were tested following Morley and colleagues [23]. Data were analysed using Graphpad Prism Version 5. Data from microarray and real time PCR populations were examined to ascertain if they fit normal distributions, using the D'Agostino and Pearson omnibus normality test and the Shapiro-Wilk normality test. Because both sets of data were not normally distributed, a Spearman correlation (Rho) was employed to test for correlation. The statistical analyses used an alpha value of 0.05.

Results/Discussion
We targeted three female tissues, namely, gastrodermis (absorptive gut lining) of the posterior halves of the worms, ovary, and the vitelline glands ( = vitellaria, accessory glands of the female system that produce precursors for eggshell synthesis) (Figures 1 and 2). We chose these three tissues due to their relative abundance, clearly delimited structure and the important biological roles in schistosome development and reproductive biology.
In view of the closely knit organization of schistosome tissues, it was important to know whether the three tissues under investigation represented homogenous cell populations. Ultrastructural assessment indicated that the ovary and gastrodermis were homogenous ( Figure 3). We had previously shown through ultrastructural studies incorporating a stereological analysis of the relative volumes of tissues in vitellogenic regions that although some parenchymal tissues intrude into the vitelline regions,     [24] which are highly synthetic cells. Thus, all tissue extracts represent homogenous or near homogenous samples.
For microarray analysis, unfixed frozen females were sectioned by cryostat onto membrane-coated slides, stained with toluidine blue and microdissected using a PALM laser catapult microscope ( Figure 1). Total RNA integrity from microdissected samples was assessed ( Figure 1) and shown to be of high fidelity. A distinct 28S band is never visible in total RNA fractions of schistosomes [25]. RNA was further processed for one-colour fluorophore-labelled cRNA synthesis and hybridization to a microarray representing the near complete transcriptome of adult schistosomes [11]. Of 38,444 probes (representing 19,222 contigs) on the chip, 8,454 (5,242 contigs) were retained after filtering (Table S4).
Principal component analysis (PCA) is a multi-dimension reduction method that allows the visual presentation of a complex data set, so that distances between plotted points represents the relative similarity of each datasets. Usually plotted in an X,Y,Z formation, each axis represents a distinct subset of data points, or in the current application, gene lists. Gene expression profiles of the three microdissected tissues and the control sample were analysed by PCA ( Figure S2). The point of the control tissue was more similar to those of the gastrodermis and vitellaria, compared with the ovary. This observation is not surprising, for the former tissues are voluminous in female parasites and likely account for much of the female transcriptome.
Complete lists of genes enriched for each tissue sample after normalization, together with lists of selected genes of interest enriched for each tissue are presented (Tables 1-3, Table S4, and Figure 2). Major gene ontologies (GOs) of differentially expressed genes for the three tissues are also shown ( Figure S1 and Table S2). Abundant transcripts enriched for each tissue encoded protein sequences for which there was little or no annotation or sequence identity.
Nine transcripts that were enriched in one of the 3 tissues were selected for validation of expression level by real time PCR using cDNA templates from the microdissected and control samples (Figure 4). Expression levels observed by real time PCR agreed with those by microarray for these genes. The microarray and real time PCR data sets of the 9 genes showed a significant correlation of 0.6791 (Spearman's Rho, p,0.0001, n = 27).
After filtering the microarray data and normalizing signal relative to female germinal tissues, we identified 214 probes representing 147 genes enriched for the gastrodermis (Figure 2, Table 1, and  Table S4). Comparable datasets in Table S4 compare gene expression of the gastrodermis relative to either ovary or vitellaria. These three datasets show strong congruence, although with some variation in relative enrichment of some sequences. Thus, a ferritin isoform is enriched in the gastrodermis relative to the ovary, but not relative to the vitellaria (Contig7767, Table S4).
The enriched genes of the gastrodermis relative to female germinal tissues included proteases of the haemoglobinolytic cascade, membrane-associated molecular transporters, actin and associated molecular motors. A highly enriched gene of the gastrodermis, represented by Contig5007, is a hitherto uncharacterized gene with uncertain sequence identity, but which contains motifs with similarity to the meprin family of metalloproteinases, and an erythrocyte-binding protein of malaria parasites. This molecule potentially represents a novel class of proteinases involved in haemoglobinolysis in these vascular parasites [26]. Surprisingly, cathepsin D, an early member of the haemoglobinolysis cascade [27], was not enriched for the gastrodermis. Given its upstream role in this multi-enzyme network, cathepsin D is probably expressed in anterior zones of the gut, either in the oesophageal gland, or in anterior zones of the gastrodermis. Our study focused on microdissection of the posterior regions of the gastrodermis. Regional specialization of the apparently simple gastrodermis of other platyhelminths has been postulated [28]. It may be that the schistosome gut displays a similar planar polarity, evidenced by distinct secretory product in different zones along the length of the parasite [26]. The hypothesis is further substantiated by observations that the gastrodermal regions analysed here were enriched for numerous sequences encoding dipeptidases and carboxypeptidases (Table  S4), peptidases more likely to be associated with terminal parts of the haemoglobinolytic cascade. Additionally, transcripts encoding proteins previously localized to the outer tegumental surface of the parasite (tetraspanins, annexin and alkaline phosphatase) were enriched for the gastrodermis relative to other tissues. Although these molecules have been previously recognised as tegumentary components, their occurrence in the syncytial gastrodermis is not surprising. Transcripts for divalent metal transporters, particularly a member of the Zinc regulated transporter/iron regulated transporter family (ZIP) family were enriched for the gastrodermis. Schistosomes have high dietary requirements for iron [29,30] and other divalent metals. While a surface mediated pathway for iron uptake by schistosomes has been postulated [29], the presence of metallotransporters in the parasite gut indicates that this tissue may also scavenge the trace metals [29].
Other transcripts enriched for the gastrodermis (relative to ovary and vitelline tissues) represent genes encoding lysosomal proteins, namely, cystinosin, lysosomal acid membrane glycoprotein (Lamp1/CD68), lysosomal alpha mannosidase and acid phosphatases ( Figure 2, Table 1, and Table S4), although lysosomes are not abundant cytoplasmic features of the gastrodermis. A consistent feature of the syncytium, however, is the presence of apical epicellular vacuoles (Figure 3), which enclose parts of ingested host blood, and are lined by villus-like lamellae. The vacuoles have many features of lysosomes, namely a low pH and presence of proteolytic enzymes [26] and are a possible cellular location for the lysosome molecules of the gastrodermis.
We identified 6,645 probes representing 4,149 upregulated genes ( Figure 2, Table 2, and Table S4) for the ovary compared with the gastrodermis. Similarly, we identified 3,832 probes representing 2,553 upregulated genes ( Figure 2, Table 3, and  Table S4) for vitellaria compared with the gastrodermis. Oocytes and vitellocytes, in platyhelminth evolution and ontogeny, are believed to be derived from common progenitor cells [31]. We decided, therefore, to determine whether the tissues have common expression identity that may reflect the common origin of the two tissues. Analysis of expression by Venn diagram indicated substantial overlap in expression between the two germinal tissues, but not with the gastrodermis ( Figure 2B). Genes enriched for both cell types included egg-specific genes including major egg antigens and egg protein cp422. The former gene is also abundant in mature eggs of schistosomes [32]. Other genes enriched for the female reproductive tract included those encoding molecules for TGF-b and tyrosine kinase signalling pathways [33,34], different innexins (gap junction proteins of invertebrates), and a diversity of genes encoding molecules associated with DNA processing, replication, and transcription. With the exception of the eggspecific antigens, the upregulated genes common to the two tissues are involved in cell proliferation and intercellular signalling. Genes enriched for the ovary (relative to the gastrodermis) included a number encoding proteins associated with cytokinesis, fertilization and coated pit-mediated endocytosis ( Table 2). Oocytes express genes with identity to polycomb, enhancers of polycomb, and Peter pan homologues of vertebrates and ecdysozoans [35] (Table 2). Polycomb genes, not previously recognized for platyhelminths, repress Hox expression in embryogenesis leading to cellular and zonal differentiation in embryos. Discovery of genes involved in embryonic differentiation will provide new insight into developmental cascades in the complex multi-generational schistosome life-cycle, leading in turn to a better understanding of differentiation of the intraovular embryo, the stage primarily responsible for pathogenesis in schistosomiasis.
Expression analysis of vitellaria (Table 3, Figure 2, and Table  S4) revealed enriched genes (relative to the gastrodermis) associated with egg-shell synthesis, as well as a range of membrane transporters with affinity for amino acids, metallo-ions and nucleotides. Eggshell precursors, egg-specific proteins and tyrosinases were upregulated as expected for this tissue that provides precursors for choriogenesis [30,36]. Numerous membranespanning transporters and genes encoding proteins for exocytosis were also enriched as were those associated with lipid metabolism. Some transcripts, annotated as containing signal peptides, did not contain abundant tyrosine residues, a prerequisite for eggshell precursors [37], possibly indicating that these molecules function in aspects other than shell formation. Given the essential role of vitellocytes in egg development and embryogenesis, functional characterization of these putative secreted proteins may enhance our understanding of the complexity of egg-shell synthesis and may help resolve long-standing questions about yolk function of vitellocytes [30].
The integration of microarray analysis of LMM-dissected tissues has provided the means to establish a gene expression atlasing strategy for S. japonicum, alleviating the technology hurdles imposed by the acoelomate nature of this platyhelminth and expediting localization of multiple genes. Tissue-specific expression profiling has been performed previously for cavitate invertebrates, incorporating LMM or gross dissection methods [38,39], but this approach demonstrates the feasibility for gene mapping in a platyhelminth, thus serving as an exemplar for similar studies of other basal bilaterians and small organisms. The localization data provided here serves as a novel resource to advance functional studies of many unannotated S. japonicum genes, thereby providing a valuable molecular platform to shed light on the complex   Relative fold change of all contigs normalised to control. Fold change for gastrodermis is normalised to the signals from an average of ovary and vitelline and individually against the two germinal tissues. The fold change for vitellaria or ovary are normalised to the signal from gastrodermis. Legend: Systematic: Probe identifiers. A full list of probes is shown in Table S1. Protein Identifier: BLASTX annotation result based on protein sequence. Normalised data and Flags are shown in separate columns for each tissue type. Fold changes relative to other tissues for the gastrodermis, ovary and vitellaria are shown is successive columns. Synonym: probe identity. Microarray Description: BLASTn annotation result based on nucleotide sequence. Abbreviations: P-Present; A-Absent; g-gastrodermis, o-ovary; v-vitellaria. In other sheets, the lists of enriched probes (Fold change greater than 2; sorted by decreasing fold changes relative to comparative tissues) for each of the three tissue types are presented on separate sheets of this file. Gastrodermis: Transcripts enriched for gastrodermis relative to ovary and vitellaria. In separate columns the fold changes of genes relative to either ovary and vitellaria are shown. Ovary: Transcripts enriched for ovary relative to gastrodermis. Vitellaria: Transcripts enriched for ovary relative to vitellaria. Legend to Sheets 2-6: Systematic: Probe identifiers. A full list of probes is shown in