Single and multi-gene phylogeny of Hepatospora (Microsporidia) – a generalist pathogen of farmed and wild crustacean hosts

Summary

Almost half of all known microsporidian taxa infect aquatic animals. Of these, many cause disease in arthropods. Hepatospora, a recently erected genus, infects epithelial cells of the hepatopancreas of wild and farmed decapod crustaceans. We isolated Hepatospora spp. from three different crustacean hosts, inhabiting different habitats and niches; marine edible crab (Cancer pagurus), estuarine and freshwater Chinese mitten crab (Eriocheir sinensis) and the marine mussel symbiont pea crab (Pinnotheres pisum). Isolates were initially compared using histology and electron microscopy revealing variation in size, polar filament arrangement and nuclear development. However, sequence analysis of the partial SSU rDNA gene could not distinguish between the isolates (~99% similarity). In an attempt to resolve the relationship between Hepatospora isolated from E. sinensis and C. pagurus, six additional gene sequences were mined from on-going unpublished genome projects (RNA polymerase, arginyl tRNA synthetase, prolyl tRNA synthetase, chitin synthase, beta tubulin and heat shock protein 70). Primers were designed based on the above gene sequences to analyse Hepatospora isolated from pea crab. Despite application of gene sequences to concatenated phylogenies, we were unable to discriminate Hepatospora isolates obtained from these hosts and concluded that they likely represent a single species or, at least subspecies thereof. In this instance, concatenated phylogenetic analysis supported the SSU-based phylogeny, and further, demonstrated that microsporidian taxonomies based upon morphology alone are unreliable, even at the level of the species. Our data, together with description of H. eriocheir in Asian crab farms, reveal a preponderance for microvariants of this parasite to infect the gut of a wide array of decapods crustacean hosts and the potential for Hepatospora to exist as a cline across wide geographies and habitats.

INTRODUCTION

Microsporidia are single-celled eukaryotic intracellular parasites known to infect a range of vertebrate and invertebrate hosts (Mathis et al. 2005). Since the inception of the phylum ‘Microsporidia’, both phylogenetic placement of the group and classification within the group have proven problematic [reviewed in (Corradi and Keeling, 2009)]. Early phylogenies used to place microsporidia within the tree of life failed to account for rate heterogeneity among gene sites, base-compositional biases and the overall accelerated evolutionary rate characteristic of microsporidian genomes (Hirt et al. 1999) and thus microsporidia were for a long-time considered basal eukaryotes. They are now, along with the Cryptomycota, considered the most basal fungi group, a new classification based on the phylogenetic analysis of 200 genes (James et al. 2013).

Within the phylum, in the past, classification of taxa has been based on structural characteristics, ultrastructural morphology and karyotypic evidence (Canning, 1953; Vavra and Undeen, 1970; Shadduck et al. 1990; Cali et al. 1993). Now there is a compounding body of evidence supporting the idea that morphological and developmental features in this phylum are plastic between both closely and distantly related microsporidia (Vossbrinck and Debrunner-Vossbrinck, 2005; Stentiford et al. 2013). Stentiford et al. (2013) observed that microsporidians isolated from marine decapod crustaceans that would have been classed as distantly related taxa (Nadelspora and Ameson) under a morphology-based classification system are in fact close relatives on rDNA-based phylogenetic trees and are potential life-cycle variants of the same taxon. An example of a highly plastic morphological character used in classification is nuclear configuration. Two configurations are known to exist in microsporidia: a monokaryon (an individual nucleus) and a diplokaryon (two apposed nuclei). Whilst some species retain the same nuclear configuration throughout their life cycle, others switch between the nuclear stages depending upon life-cycle stage, and polymorphic species shift between the nuclear stages when changing hosts or tissues (Vávra and Larsson, 2014).

For many years the alternative to morphological characters has been the SSU rRNA gene, which has now been amplified and sequenced from over 1000 species of microsporidia. However, the rDNA sequences amplified are generally only able to resolve taxonomic relationships down to the genus level (Vossbrinck et al. 1998; Vossbrinck and Debrunner-Vossbrinck, 2005). The most commonly used primers, F18 and 1492R (Vossbrinck et al. 1993; Kent et al. 1996) amplify a 1400 bp fragment of the most conserved region of the SSU rDNA and omit the internal transcribed spacer (ITS) (Vossbrinck and Debrunner-Vossbrinck, 2005), a region reported to be highly variable even within species and hence highly informative for intra-species differentiation analyses (Gresoviac et al. 2000; Sak et al. 2011). However, even if this region was commonly amplified for microsporidia, there are other issues with using SSU as a marker for closely related species. In at least some microsporidia, the process of concerted evolution that typically keeps rDNA copies uniform within the genome, does not seem to be present. For example, Ironside (2013), O'Mahony et al. (2007) and Tay et al. (2005) observed a higher ITS sequence variation between repeats in the same genome than between different genomes for some Nosema species, which means that caution needs to be used when employing rDNA to discriminate between closely related microsporidia.

One of the issues emerging from the difficulty in resolving close phylogenetic relationships and morphological plasticity in microsporidians is the assignment of appropriate species names. This is an important issue because taxonomic names of pathogenic species are fed into legislative frameworks that are used to inform policy making (Stentiford et al. 2014). Therefore there is a growing need to find alternative molecular markers to resolve questions in microsporidia taxonomy. There is currently whole-genome data for 20 microsporidian species on the publicly available MicrosporidiaDB database. With cheaper thorough-put sequencing technologies and the advent of single cell genome sequencing, we can only expect this number to increase. This provides a minable resource for molecular characters for multi-locus phylogenies (Capella-gutiérrez et al. 2012).

We isolated putative Hepatospora sp. parasites from three different crustacean hosts, inhabiting different habitats and niches; marine edible crab (Cancer pagurus), estuarine and freshwater Chinese mitten crab (Eriocheir sinensis) and the marine mussel symbiont pea crab (Pinnotheres pisum). Isolates were initially compared using histology and electron microscopy revealing not only a similar life cycle, but also the variation in size, polar filament arrangement and nuclear development (Table 1). However, sequence analysis of the partial SSU rDNA gene could not distinguish between the isolates (≈99% similarity). Here we take advantage of the current microsporidian whole-genome database and our in-house genomic data to construct a six-gene concatenated phylogenetic tree for Hepatospora eriocheir, a parasite of the invasive Chinese mitten crab (Stentiford et al. 2011), a pea crab (P. pisum) infecting microsporidium (Longshaw et al. 2012) and a novel microsporidium that infects commercially important edible crabs (C. pagurus), all from European waters. Our results revealed that Hepatospora isolates from the three different crustacean hosts are likely to be microvariants of a single species. Further, they support the concept that microsporidian taxonomies based upon morphology are not only unreliable, but can also be deceptive, even at the level of the species. Our data, together with description of H. eriocheir as an agent of emergent disease in Asian crab aquaculture (Ding et al. 2016) reveal a preponderance for microvariants of H. eriocheir to infect the gut of a wide array of decapods crustacean hosts across wide geographic boundaries and in a range of habitats.

Table 1. Comparison of host and structure [histological, electron microscopy and molecular data (SSU 18s rRNA)] from the Hepatospora isolates

MATERIALS AND METHODS

Eriocheir sinensis, Cancer pagurus and Pinnotheres pisum sampling

Chinese mitten crabs (E. sinensis) were sampled from two locations in the Thames Estuary; a site near to the Millennium Dome (51:27:12N, 00-00-44E) and another at Tilbury Power Station (51:27:12N, 00-23-10E). Crabs were collected using Fyke nets and from the screens of water intake pipes at the power station. Edible crabs (C. pagurus) were captured using baited pots in the Weymouth and Portland area of the English Channel, UK (50°32′50″N, 002°11′00″W) as previously described (Bateman et al. 2011). Live crabs were transported to the Cefas laboratory in Weymouth and anaesthetized by chilling on ice for 30 min before dissection. The hepatopancreas, gill, gonad, central nerve ganglia, heart and body muscle was removed from the crabs and fixed in Davidson's sea water fixative for histology. Additional hepatopancreas samples were fixed in 2·5% glutaraldehyde in 0·1 m sodium cacodylate buffer for electron microscopy, and 100% ethanol and frozen, for molecular analyses.

For P. pisum sampling, blue mussels (Mytilus edulis) were collected from a range of sites around the UK. Mussels were opened by cutting the abductor muscle and separating the valves. Symbiotic pea crabs (P. pisum) found habituating the mantle of the host mussel were removed and anaesthetized on ice prior to bilateral dissection and removal of the hepatopancreas. Small portions of the organ were fixed in Davidson's sea water fixative, 2·5% glutaraldehyde in 0·1 m sodium cacodylate buffer and 100% ethanol for histology, electron microscopy and molecular techniques, respectively.

Histology and transmission electron microscopy

For histology, fixation was allowed to proceed for 24 h before samples were transferred to 70% industrial methylated spirit. Fixed samples were processed to wax in a vacuum infiltration processor using standard protocols. Sections were cut at a thickness of 3–5 µm on a rotary microtome and mounted onto glass slides before staining with haematoxylin and eosin (H&E) and Feulgen stains. Stained sections were analysed by light microscopy (Nikon Eclipse E800) and digital images and measurements were taken using the LuciaTM Screen Measurement System (Nikon, UK). For electron microscopy, tissues were fixed in 2·5% glutaraldehyde in 0·1 m sodium cacodylate buffer (pH 7·4) for 2 h at room temperature and rinsed in 0·1 m sodium cacodylate buffer (pH 7·4). Tissues were post-fixed for 1 h in 1% osmium tetroxide in 0·1 m sodium cacodylate buffer. Samples were washed in three changes of 0·1 m sodium cacodylate buffer before dehydration through a graded acetone series. Samples were embedded in Agar 100 epoxy (Agar Scientific, Agar 100 pre-mix kit medium) and polymerized overnight at 60 °C in an oven. Semi-thin (1–2 µm) sections were stained with Toluidine Blue for viewing with a light microscope to identify suitable target areas. Ultrathin sections (70–90 nm) of these areas were mounted on uncoated copper grids and stained with 2% aqueous uranyl acetate and Reynolds’ lead citrate (Reynolds, 1963). Grids were examined using a JEOL JEM 1400 transmission electron microscope and digital images captured using an AMT XR80 camera and AMTv602 software.

Spore isolation and DNA extraction

Hepatopancreas samples from histologically confirmed microsporidian-infected crabs were crushed with a sterile pestle and mortar in PBS. The homogenous mash was filtered through a 100 µm mesh followed by cell sieving through 40 µm filter. Filtrate was topped up to 50 mL using PBS/triton-X (0·1%) and pelleted at 3220 g in an Eppendorf centrifuge precooled at 4 °C, for 10 min. The supernatant was removed and the pellet (containing host cell debris and microsporidian spores) was resuspended in 2·5 mL of ice-cold water. Homogenate was added to the top of a Percoll density gradient and centrifuged at 1000 g in a pre-cooled centrifuge for 45 min. Phase contrast and fluorescence microscopy were used to visualize the spores and to examine for purity following Percoll gradient purification. Spores were stained with the chitin marker calcoflour white (Darken, 1962).

For extraction of DNA, spores of the Pea crab and Mitten crab parasites were diluted 1 in 10 (w/v) in G2 buffer and Proteinase K was added at a concentration of 2 mg mL⁻¹ (Qiagen, UK). The spores were subsequently disrupted in a Matrix D FastPrep cell disrupter (FastPrep, UK) by shaking on a homogenizer for 2 min at highest setting. Homogenized samples were incubated for 4 h at 56 °C. Total DNA was extracted using an EZ1 DNA tissue kit and EZ1 Advanced XL BioRobot (Qiagen) following manufacturers’ instructions. For the edible crab parasite, aliquots of purified spores suspended in 1 × PBS were mixed with liquid nitrogen in a sterile mortar. The mixture was slowly stirred until it solidified, after which it was ground with a sterile pestle for 10 min. Liquid nitrogen was again added to the powder and the mix was ground for an additional 10 min. This step was repeated three times before dissolving the resulting powder in 800 µL of phenol (pH 8·0). The homogenate was transferred to an Eppendorf tube and mixed by inversion and subsequently centrifuged for 10 min at 10 000 × g. The recovered supernatant was mixed by inversion with 400 µL of chloroform and centrifuged for another 10 min at 10 000 × g. Genomic DNA was precipitated from the aqueous solution using a standard ethanol precipitation protocol (Ausubel et al. 2002) and sent to the University of Exeter sequencing service, UK for library preparation and Illumina sequencing.

Identification of six marker genes from unpublished genome projects of H. eriocheir and the edible crab microsporidian

Orthologues of the six marker genes [amino acyl tRNA synthetases (Brown and Doolittle, 1999): arginyl tRNA synthetase and prolyl tRNA synthetase, beta-tubulin (Edlind et al. 1994), chitin synthase (Hinkle et al. 1997), heat shock protein 70 (HSP70) (Hirt et al. 1997) and RNA polymerase II (Hirt et al. 1999)] for publicly available microsporidian genomes were obtained from the MicrosporidiaDB database (Aurrecoechea et al. 2011) by initially performing word searches for the individual marker gene names for Encephalitozoon cuniculi GB-M1 and then using the ‘transform by orthology tool’ to find orthologues for all the other microsporidians in the database. To identify the desired marker genes in the newly sequenced genomes of H. eriocheir and the edible crab microsporidian, predicted open reading frames (ORFs) were queried using the microsporidian proteins obtained from MicrosporidiaDB (Aurrecoechea et al. 2011) using command line blastn (Mount, 2007) with an e-value cutoff of 1e-5. The top ORF hits were selected as the orthologous genes and orthology was verified by assessment of phylogenies of single genes.

Primer design, PCR and sequencing of the six-marker gene from the pea crab parasite

Due to the small amount of pea crab parasite genomic DNA recovered from the extraction procedure, gene-specific PCRs and subsequent sequencing was performed to retrieve the corresponding sequences for the six genes of the pea crab parasite rather than full genome sequencing. We designed gene-specific primers by using the first and last 18 nucleotides of the selected orthologues from H. eriocheir. A two-step nested PCR was done to amplify longer genes such as Arginyl tRNA and RNA polymerase (Table 2).

Table 2. Gene-specific primers were designed using the first and last 18 nucleotides of the selected orthologues from H. eriocheir

Ribosomal DNA phylogenetic analyses

Universal primers were used to amplify rDNA regions from genomic material extracted from H. eriocheir, pea crab and edible crab parasites (Table 3). The sequencing results for the PCR products were subsequently aligned and used to construct a neighbour-joining tree with MEGA software (Tamura et al. 2011). All PCR reactions were performed in a 50 µL reaction mix consisting of 1× Green Go Taq buffer, 2·5 mm MgCl₂, 0·25 mm dNTPs, 100 pm each of the forward and reverse primer sets, 0·25 units Go Taq Flexi (Promega, UK) and 2·5 µL of extracted genomic DNA. Amplifications were performed on a Peltier PTC-225 thermal cycler with the following settings: Initialization step at 94 °C for 5 min, 40 cycles of 1 min denaturation at 95 °C, a 1 min annealing step (see Tables 1 and 2 for temperatures) and a 1 min extension step at 72 °C. A final elongation step was carried out for 10 min at 72 °C. Amplification products were resolved on 2% agarose gels stained with ethidium bromide and visualized using a UV illuminator.

Table 3. rDNA primers used in this study

Correct size products (Tables 2 and 3) were excised from the gels and purified using the Wizard SV gel and PCR purification system (Promega, UK). PCR products were sequenced using Sanger technology, ABI PRISIM Big Dye Terminator v3·1 cycle sequencing kit (Thermo Fisher Scientific, UK) with relevant primers (Tables 2 and 3) and the following PCR thermal cycling program: 94 °C × 30 s followed by 30 cycles of 96 °C × 10 s, 50 °C × 10 s and 60 °C × 4 min and held at 4 °C. Analyses of the sequenced PCR product were done using the Sequencher software (Gene codes corporation).

Construction of a six-gene phylogenetic tree

Maximum likelihood (ML) analyses

The orthologues for each gene set were individually aligned with the command line MUSCLE program (v3·8·31) (Edgar, 2004) using the default settings, and then subsequently masked with the automatic command line tool TrimAl (Capella-Gutierrez et al. 2009). Saccharomyces cerevisiae was used as an out-group in each of the microsporidian datasets.

A GTR substitution model with a gamma model of rate heterogeneity was used to create ML trees for the individual gene sets with the RaxML program (Stamatakis, 2014). These were pilot trees to check for unusually long-branch lengths indicative of unlikely orthologues. The masked genes from each microsporidian species were subsequently manually concatenated using SeaView (v4) (Gouy et al. 2010). For the final construction of the ML concatenated gene tree, a partition file that contained the positions of the individual genes within the alignment was manually created and passed to the RaxML program using the ‘–q’ option (Stamatakis, 2014). This was to enable the program to treat each gene set in the concatenated alignment separately and allow it to estimate individual nucleotide substitution rates. These estimations were also performed with the GTR + GAMMA nucleotide substitution model.

Bayesian inference analysis on six-gene concatenated alignment

To check for reliability of the phylogenetic relationships estimated by ML analyses, a Bayesian inference method was also used to reconstruct the six-gene concatenated phylogenetic tree using MrBayes program (v3·2) (Ronquist et al. 2012). A partition file containing positions of the individual genes in the alignment was created according to the program manual. The program was run using a GTR + GAMMA model and probability distributions were generated using the Markov Chain Monte Carlo Methods. A total of 1 020 000 generations were run, the first 25% of sampled trees were discarded as ‘burn-in’ and a consensus tree was constructed.

RESULTS

Pathology, ultrastructure and SSU-based phylogeny

Infected hepatopancreatic tubules from all three crab hosts displayed varying proportions of infected epithelial cells consistent with previous descriptions of Hepatospora-associated pathology (Stentiford et al. 2011). Infection was confined to the epithelial cells of the hepatopancreas, with no other organ seemingly infected. However, in some severe cases where hepatopancreatic tubules were disrupted, liberated parasite spores could be observed within the lumen of affected tubules and, in the haemolymph. Histologically, the disease caused by this microsporidian was indistinguishable between Chinese mitten crab (where the parasite has been confirmed as H. eriocheir), edible crab and pea crab (Fig. 1).

Fig. 1. (A) Histopathology of Hepatospora sp. in the hepatopancreas of E. sinensis. Scale = 50 µm. (B) Histopathology of H. eriocheir in the hepatopancreas of Cancer pagurus. Scale = 100 µm. (C) Histopathology of Hepatospora sp. in the hepatopancreas of P. pisum. Scale = 25 µm. Tubule epithelial cells contained multiple granular inclusions (arrows) All images H&E histology.

Ultrastructural analysis revealed multiple stages of a microsporidian parasite within the cytoplasm of hepatopancreatic epithelial cells of each host crab species. The earliest stage observed was the meront (Fig 2A, C and E). Hepatospora eriocheir (infecting Chinese mitten crab) possessed uni-nucleate meronts, whereas the parasite infecting pea crab and edible crab possessed bi-nucleate meronts. A similar observation was made in the spore stages with spores (Fig 2B, D and F) from the parasite infecting edible crab and pea crab appearing bi-nucleate, while spores from the Chinese mitten crab were of uni-nucleate karyotype. Mature spores in all cases possessed a trilaminar wall consisting of a plasma membrane, an electron lucent endospore and an electron-dense exospore. Spores measured 1·8–1·9 × 0·9 µm but contained varying turns of an isofilar polar filament. Spores from the parasite-infecting Chinese mitten crabs and edible crabs possessed seven to eight turns of the polar filament while those from the parasite-infecting pea crabs possessed five to six turns of the polar filament. A summary of these shared and distinctive features is given in Table 1.

Fig. 2. Electron micrographs of H. eriocheir from E. sinensis (A and B) and Hepatospora sp. from Cancer pagurus (C and D) and P. pisum (E and F). (A) Developing sporonts within a parasitophorous vesicle. Sporonts contain multiple unikaryotic nuclei (arrows) which then divide and mature to form spores. Scale bar = 500 nm. (B) Mature spore containing seven to eight turns of the polar filament (arrow) in single file, unikaryotic nucleus (uni), anchoring disc (a) and polaroplast (p). Scale bar = 100 nm. (C) Developing sporonts within a parasitophorous vesicle. Sporonts contain multiple diplokaryotic nuclei (*) which then divide and mature to form spores. Scale bar = 0·5 µm. (D) Mature spore containing five to six turns of the polar filament (arrow) in single file, diplokaryotic nucleus (dip), anchoring disc (a) and polaroplast (p). Scale bar = 0·2 µm. (E) Developing sporonts within a parasitophorous vesicle. Sporonts contain multiple diplokaryotic nuclei (*) which then divide and mature to form spores. Scale bar = 500 nm. (F) Mature spore containing five to six turns of the polar filament (arrow) in single file, diplokaryotic nucleus (dip), anchoring disc (a) and polaroplast (p). Scale bar = 100 nm.

Partial sequencing of the SSU rDNA gene obtained from the parasites infecting the three host crab species revealed an apparent synonymy (≈99% similarity over 890 bp), at least based upon this portion of the SSU, between the three parasites. Despite some ultrastructural and karytopic distinctions between the three isolates (Table 1), SSU-based phylogeny did not support erection of distinctive taxa for the parasites infecting edible crab and pea crab. Based upon SSU phylogeny, these parasites would therefore be classified as H. eriocheir, with edible crab and pea crab representing an extended host range for this parasite.

Phylogeny of Hepatospora isolates based upon six concatenated genes

Taking in to account the potential weakness of SSU-based phylogenies for discriminating closely related microsporidian (and other) taxa, phylogenies based upon alternative (coding) regions of the Hepatospora genome were constructed to investigate their potential as taxonomic discriminators (GenBank accession nos. of Hepatospora genes used: KU695715, KU695716, KU695717, KU695718, KU695719, KU695720, KU695721, KU695722, KU695723, KU695724, KU695725, KU695726, KU695727, KU695728, KU695729, KU695730, KU695731, KU695732). The resulting alignment of the six masked concatenated genes consisted of 18 232 sites. Phylogenetic trees based on ML and BI methods displayed identical topologies and strongly supported the grouping of the Chinese mitten crab parasite H. eriocheir, the edible crab and pea crab parasites, and the placement of E. cuniculi strains and Nematocida spp. as distinct clades with high confidence values (Fig. 4). The clade consisting of H. eriocheir, the edible crab parasite and the pea crab parasites branched as a sister group to Enterocytozoon bieneusi as consistent with previously phylogenies based upon SSU gene sequences (Stentiford et al. 2011). Our tree is also consistent with a previous multi-protein microsporidian phylogeny showing the Enterocytozoonidae forming a clade with Vittaforma corneae and with the Nosema/Encephalitozoon clade as a sister group to these (Nakjang et al. 2013). Within our tree, all resequenced strains are retrieved as clades with few nucleotide differences separating them. As observed for our crab pathogens, no nucleotide differences separated strains of Nematocida sp. or Nematocida parisii, in our analysis. Nucleotide differences do however separate E. cuniculi EC1-3 strains from E. cuniculi GB and E. hellem swiss from E. hellem ATCC. The six gene concatenated phylogeny supports the proposition from our SSU-based phylogeny that the parasites infecting Chinese mitten crabs, European edible crabs and, pea crabs can all be classified as H. eriocheir. Further, it provides firm evidence that H. eriocheir (or very closely related microvariants thereof) can infect a wide variety of decapods crustaceans from different aquatic habitats.

DISCUSSION

Phylogenetic analyses performed with rDNA have been pivotal in erecting new taxa and in the discovery of morphological plasticity within the microsporidian phylum but have however been unsuccessful in resolving the branching relationships between closely related species. The Hepatospora genus is a recent taxa erected as a result of modern phylogenetic techniques (Stentiford et al. 2011). It has been defined as a genus that encompasses microsporidia with infective life stages that develop within the hepatopancreas of marine and brackish water hosts. With the use of rDNA-based phylogenetic analyses, Stentiford et al. (2011) renamed and updated the description of Endoreticulatus eriocheir (Wang and Chen, 2007) to form the type species H. eriocheir. Unlike Wang and Chen who isolated their spores from native Chinese mitten crabs, the spores in the Stentiford et al. (2011) study were isolated from invasive Chinese mitten crabs (E. sinensis) which had been caught in the Thames estuary, UK. In both studies, the spores were described as ellipsoid, measuring ~1·8 × 0·9 µm and containing seven to eight polar filaments. Recently, Ding et al. (2016) have shown that not only is the pathogen infecting the Asian population of Chinese mitten crabs H. eriocheir, but also the infection is associated with an emerging disease condition causing significant proportions in aquaculture production of this species in China. In their original taxonomic paper, Stentiford et al. (2011) also highlighted an unassigned microsporidian infecting the hepatopancreas of edible crabs, C. pagurus, and suggested at the time that this was also likely a member of the genus Hepatospora based upon high similarity with the partial SSU gene sequence from H. eriocheir.

In 2012, Longshaw et al. produced a disease profile of the pea crab (P. pisum) revealing the presence of two uncategorized microsporidian parasites. One of these appeared to be cytoplasmic, residing in hepatopancreatocytes and inducing a necrotizing effect that resulted in the degeneration of hepatopancreatic tubules. However, in this study, the microsporidian was not assigned a taxon and no further data was presented in order to aid its molecular or morphological characterization.

In the present study, we have shown that infection with the parasite of the hepatopancreas of edible crabs displays a similar pathology to that caused by infection with H. eriocheir in Chinese mitten crabs from London (Stentiford et al. 2011) and from China (Wang and Chen, 2007; Ding et al. 2016), and to the infection described in pea crabs by Longshaw et al. (2012). These parasites also appear to share a broadly similar morphological development within host gut epithelial cells albeit with some distinctive differences in karyo-status (H. eriocheir is unikaryotic whereas the pea crab and edible crab parasites are dikaryotic) and some minor differences in polar tube-coiling patterns (summarized in Table 1). Based upon our analysis of rDNA gene sequence-based phylogenetic relationship between H. eriocheir and the two novel parasites, all three are virtually indistinguishable, forming a monophyletic group immediately adjacent to the Enterocytozoonidae clade (Stentiford et al. 2013) (see Fig. 3).

Fig. 3. Neighbour – joining tree based on a 511 bp nucleotide partial SSU 18s sequence from parasites isolated from hepatopancreas of E. sinensis (H. eriocheir), Cancer pagurus (Hepatospora sp.) and P. pisum. The phylogenetic analysis was performed using Mega version 5·05. Analysis was done using 1000 bootstrapped datasets and values >70% are shown on the tree. The scale bar represents substitutions per nucleotide site.

Based upon the somewhat surprising finding that the same parasite taxon appeared to infect not only Chinese mitten crabs from different parts of their invasive/native range, but also occurred in niche-separated crab hosts from the European marine environment we used these three microsporidians as a model system to assess how a concatenated multi-gene phylogenetic approach could be applied as refined tool for discriminating taxa which cannot be separated by rDNA-based phylogenetic approaches. As outlined in our recent Opinion piece, appropriate application of such techniques may be suitable where significant biological reasoning exists for potential separation of taxa (e.g. one causing disease and another not) and, may be particularly important where pathogens are to be listed to prevent their international trade in animals and their products (Stentiford et al. 2014). The model system presented by Hepatospora was appropriate given the potential for geographical, host and habitat distinction between known isolates, some reasonably significant differences in morphology (i.e. karyotstatus) and, the availability of draft genome data for the best studied of these, H. eriocheir. The rationale behind this concatenated phylogeny approach was to look for alternatives or additional data with which to resolve the branching relationship between very closely related microsporidia. Despite this reasoning, our concatenated phylogenies resulted in almost identical tree topologies in both the ML and BI probabilistic approaches used with nodes supported by high bootstrap and posterior probability values, respectively. This, in addition to the retrieval of well-known relationships like the grouping of Encephalitozoon and Nematocida strains (in Fig. 4) and overall tree topology similar to previously published work based on rDNA genes (Troemel et al. 2008; Stentiford et al. 2011; Cuomo et al. 2012) increases our confidence in the phylogenetic relationships inferred by our study.

Fig. 4. Grouping of three Hepatospora/Hepatospora-like species suggests they are closely related. Phylogenetic trees based on (A.) maximum likelihood, (B.) Bayesian inference of 20 microsporidians for six concatenated genes rooted with S. cerevisiae. Numbers on nodes are (A.) Bootstrap confidence levels from 100 replicates, (B.) Bayesian posterior probability values. Both trees displaying identical topologies and grouping of Hepatospora/Hepatospora-like clade are shown in bold. The scale bars represent nucleotide substitutions per site.

Whilst our multi-gene phylogeny confirms the close relationship of the three isolates, it does not discriminate well between them. This is in part due to the use of highly conserved genes for our concatenated tree phylogenetic analyses. Even though protein-coding genes have been successfully used in previous studies to infer deep phylogenetic relationships within the microsporidian phylum and placing microsporidians within the tree of life (Fast et al. 1999; Hirt et al. 1999; Keeling, 2003; Nakjang et al. 2013), the selected genes were unable to properly resolve branching relationships in our analyses due to the high level of nucleotide sequence similarity between the three Hepatospora strains (Table 1 and Fig. 3). Due to the small size of the pea crab hepatopancreas and very low infection levels in this host, we were unable to extract a large quantity of parasite genomic DNA. This in turn limited the number of marker genes we could amplify from the pea crab parasite for this study. Given more material we would have amplified further genes, for example polar tube and spore wall protein-coding genes. These have been used in previous studies to distinguish between Encephalitozoon and Nosema strains (Peuvel et al. 2000; Polonais et al. 2010; Chaimanee et al. 2011), and could have been added in our analyses to improve the resolution of branching relationships between the Hepatospora strains. It must however be noted that even though Polonais et al. (2010) were successful in differentiating between Encephalitozoon hellem strains by looking at nucleotide polymorphisms at the spore wall protein gene, this gene was unsuccessful in differentiating between Nosema ceranae geographical isolates in similar recent studies (Roudel et al. 2013; Van der Zee et al. 2014). This highlights the different evolutionary pressures acting on the genes and possibly genomes of different microsporidians and that a gene successfully used to differentiate between strains in one microsporidian species may not be ideal for other species.

Future studies of this kind should probably focus on divergent single copy orthologues between strains of interest. Primer design for more variable genes is more challenging, but the advent of single-cell sequencing will remove this issue completely. As microsporidian genomes become increasingly available, we envisage a switch from phylogenetics to phylogenomics as the latter presents a more holistic approach to understanding close phylogenetic relationships and providing information for more robust taxonomic assignments.

Taxonomy of Hepatospora

Despite the caveats exposed in the preceding section, it is useful to consider the taxonomic placement of the parasites infecting the three host crabs. Here, it is useful to look to other genera within the phylum. Since already established subspecies of E. cuniculi, E. hellem and Nematocida (also included within our analysis) have nucleotide similarity for the same genes used in this study ranging between 92·13 and 99·95%, assigning different species names to the host-specific isolates of Hepatospora (98·53–98·88% nucleotide similarity) is not supported (see highlighted boxes in Table 4). On the bases of their high nucleotide similarity and their consequent grouping with minimal branching distances on our multi-gene phylogenetic tree (shown in bold in Fig. 4), we propose that Hepatospora parasites investigated in the current study should therefore be regarded as the same species, H. eriocheir, or potentially very closely related microvariants thereof. However, in consideration of the differences in morphological features that would have placed these microsporidians in completely different taxonomic ranks using traditional approaches (and considering that taxonomy of the phylum abides by rules laid down by the ICZN), it is perhaps appropriate to consider these microvariants as subspecies. We therefore propose the assignment of H. eriocheir pinnotheres and H. eriocheir canceri as subspecies of H. eriocheir infecting the pea crab and edible crab, respectively. Regardless of use of particular nomenclature, it is noteworthy that the genus Hepatospora (and perhaps specifically its type taxon H. eriocheir) may represent an example of a parasite cline, infecting the guts of one of the most abundant host groups in our oceans, the crustaceans. The minor differences in rDNA-based and even concatenated phylogenies for Hepatospora may underlie a subtly shifting genome required for survival in hosts from different habitats. As more subspecies are discovered from variant hosts in different habitats, the concept of the parasite cline can be better studied.

Table 4. High nucleotide sequence similarity of the six marker genes used in this study between the parasites isolated from three different crab hosts

Numbers in bold are percentage identity comparison of sites (i.e. nucleotides + gaps) resulting from the alignment of each of the six marker genes of three Hepatospora/Hepatospora-like microsporidian species. Italicized numbers represent number of variable nucleotides in the pairwise alignment of two species without taking gaps into account (number given of variable nucleotides/total number of aligned nucleotides).

An alternative hypothesis would reflect the potential that a previously host-specific parasite, H. eriocheir has been inadvertently introduced to European waters by its invasive host, the Chinese mitten crab; following which it has subsequently switched to hosts which at least have niche overlap at some point in their life cycle (Ingle, 1980; Lawton, 1989; Clark et al. 1998; Becker and Türkay, 2010). The description of gut infecting microsporidian taxa from locations where such niche overlap are absent will address this key question.