Species complexes and phylogenetic lineages of Hoferellus (Myxozoa, Cnidaria) including revision of the genus: A problematic case for taxonomy

Background Myxozoans are metazoan parasites whose traditional spore morphology-based taxonomy conflicts DNA based phylogenies. Freshwater species of the genus Hoferellus are parasites of the excretory system, with several members infecting food and ornamental fish species, as well as amphibians. This study aims to increase our understanding of their molecular diversity and development, aspects about which little is known, and to generate a molecular diagnostic tool to discriminate between different pathogenic and non-pathogenic Hoferellus spp. Methods SSU and ITS rDNA phylogeny, along with morphological descriptions using light and electron microscopy were used to identify and characterize Hoferellus species collected from the urinary system of fishes and frogs. A PCR-based diagnostic assay was designed to differentiate between cryptic Hoferellus spp in cyprinid fishes commonly cultured in Central Europe. Results Our phylogenetic results separate the species of Hoferellus into two phylogenetic sublineages which are indistinguishable on the basis of generic morphological traits: 1) The Hoferellus sensu stricto sublineage, which is composed of the type species Hoferellus cyprini, Hoferellus carassii and a cryptic species, Hoferellus sp. detected only molecularly in common carp. 2) The Hoferellus sensu lato sublineage into which the new species we described in this study, Hoferellus gnathonemi sp. n. from the kidney of the elephantnose fish and Hoferellus anurae from reed frogs, are placed together with Hoferellus gilsoni previously sequenced from European eel. Apart from phylogenetic analyses, we also provide novel ultrastructural data on the phagocytotic nature of some Hoferellus plasmodia and on the elusive intracellular stages ascribed to the presporogonic development of this genus. Conclusions We provide molecular evidence of the polyphyly of the genus Hoferellus and provide novel morphological details of its members. Based on the presented data, we revise and propose emendation of the genus Hoferellus.


Background
Myxozoans are microscopic cnidarian parasites in aquatic environments and are known for the diseases they provoke in fisheries and aquaculture. They have complex life cycles alternating between intermediate vertebrate host, usually fish but also other vertebrates, and a definitive invertebrate host, annelids and bryozoans. Myxozoa have notoriously conflicting traditional morphology-and more recent phylogeny-based systematics. As a result, a paraphyletic and/or polyphyletic nature has been revealed in many myxozoan genera subjected to extensive sampling and subsequent phylogenetic analyses [1][2][3][4].
The urinary system of the vertebrate host is a common habitat for proliferation and spore formation of diverse myxozoan species [5,6]. The common myxozoan ancestor was proposed to infect renal tubules in freshwater fish, a habitat that has been independently colonized by myxozoans several times during evolution [7]. Members of the genus Hoferellus Berg, 1898; syn. Hoferia Doflein, 1898 (hom. Hoferia Bittner, 1894), syn. Mitraspora Fujita, 1912 are parasites of the excretory system, including 25 nominal species infecting fish and one species infecting frogs (see review [5]). Myxospores of Hoferellus are miterlike or round with characteristic posterior filaments or a brush border at the posterior end. Based solely on morphological data, Hoferellus has suffered abundant taxonomic reassignments [8][9][10][11][12][13]. The type species, Hoferellus cyprini (Doflein, 1898) affects the renal system of common carp Cyprinus carpio L., one of the most cultured freshwater fish worldwide [14]. The integrity of H. cyprini as a species has been discussed due to elusive intracellular stages in the epithelium of the renal tubules and the difficulty of detecting spores [15,16]. The most studied species, Hoferellus carassii Achmerov, 1960, originally described in gibel carp Carassius gibelio (Bloch, 1782), was identified as an agent of kidney enlargement disease (KED) in goldfish, Carassius auratus (L.), perhaps globally the most widespread ornamental fish, and later related to kidney bloater in farmed C. auratus, causing a polycystic, swollen kidney and consequent abdominal distension [17][18][19][20][21]. The identity, host specificity and putative occurrence of mixed infections of H. carassii and H. cyprini has been largely discussed and remains a puzzle [9,12,17,19,[22][23][24][25][26]. Most other species are known solely based on morphological description of spores, and only scarce molecular information is available for the members of this genus (2 partial SSU rDNA sequences).
An annual life cycle was assigned to H. carassii and H. cyprini [17,20,27,28]. The developmental cycle includes intracellular stages in the epithelium of the renal tubules [20,27] that were not detected by all authors [25] or ascribed to another myxozoan genus, Sphaerospora, which is expected to be unable to complete its development in the kidney [16,29]. Luminal stages of Hoferellus gilsoni (Debaisieux, 1925) are characterized by a mode of attachment in the urinary bladder of Anguilla anguilla (L.), which is unique amongst Myxozoa, including desmosome-like zones and digestive vacuoles probably related to digestion of host cell components [30].
Despite the importance of some species of Hoferellus as agents of disease in cultured fish, comprehensive data on their development, their molecular diversity and diagnostic tools for the differentiation of pathogenic vs non-pathogenic members are inexistent. During our parasitological study of fish and anuran urinary systems, different Hoferellus spp. were detected. The main goal of this study was to obtain SSU and ITS rDNA sequences to develop a molecular diagnostic assay differentiating Hoferellus spp. but also to determine the phylogenetic relationships between the members of this genus, and clarify the Hoferellus species complex infecting cyprinids in the Central European extensive aquaculture. We also provide novel ultrastructural details of plasmodia, evidencing the putatively phagocytic nature of their surface. As a result of the present study, a review of the genus Hoferellus is provided, together with its taxonomical emendation in the light of the newly obtained data.

Host sampling sites and collection methods
Cyprinid fish were obtained from eight different localities, ponds and farms in the region of South Bohemia, Czech Republic, between February and November 2011-2013 (Table 1): Common carp C. carpio (n = 131; total length 2-50 cm, weight 1-2500 g), goldfish C. auratus (n = 114; total length 2-20 cm, weight 0.6-150 g) and Prussian carp C. gibelio (Bloch, 1782) (n = 12; total length 10.2-22.5 cm, weight 6-210 g). Peters' elephantnose fish, Gnathonemus petersii (Günther, 1862) (n = 10; total length 8-12.5 cm, weight 2.8-9.75 g) imported from Nigeria (Africa), was obtained from a pet shop. All fish were transported alive to the Laboratory of Fish Protistology, at the Institute of Parasitology, euthanized by stunning (60 ppm of clove oil overdose) followed by neural pithing. Urinary system, including kidney, ureters (only in large fishes), and urinary bladder of each fish was dissected. In order to avoid DNA cross contamination, 10 % hydrogen peroxide was used routinely to clean scissors and tweezers during sampling.
A total of 17 adult male reed frogs of the genus Hyperolius (Hyperoliidae) were examined: Hyperolius kivuensis Ahl, 1931 (n = 7), Hyperolius viridiflavus (Duméril et Bibron, 1841) (n = 10). All frogs were collected in November 2010 in/around a small roadside pond~420 m south-east of Udo's campsite, Kakamega forest reserve, Kenya; 0°20'56"N, 34°51'55.9"E, 1600 m a.s.l. Frogs were euthanized by pithing and examined for myxosporeans within 24 h after collection. Prior to dissection, dorsal-, ventral-and lateral-view photographs of live and/or freshly dead, individual frogs were taken for identification purposes. Samples (1-2 mm each) of kidneys were dissected and preserved in 10 % buffered formalin and absolute ethanol, stored for several months and processed routinely for histology and DNA sequencing, respectively. Vouchers of processed frogs were deposited in the herpetological section of the National Museums of Kenya in Nairobi under accession numbers A5264/1-7 (H. kivuensis), A5265/1-10 (H. viridiflavus).
All animal procedures were performed in accordance with Czech legislation (section 29 of Act No.246/1992 Coll. on protection of animals against cruelty, as amended by Act No. 77/2004 Coll.). We declare that animal handling complied with the relevant European and international guidelines on animal welfare, namely Directive 2010/63/EU on the protection of animals used for scientific purposes and the guidelines and recommendations of the Federation of Laboratory Animal Science Associations. Permit Table 1 Hosts analyzed within the scope of this study between 2011-2013, comparative prevalence of Hoferellus spp. by species and locality as revealed by light microscopy and by single/duplex PCR screening (SSU, ITS) of kidney, ureters and/or urinary bladder and sequence data obtained in this study. All sampled cyprinids originated from the Czech Republic; Gnathonemus petersii originated from pet shop (imported from Nigeria); frogs (Hyperolius spp.) were sampled separately in 2010 in Kenya petersii and 10 plasmodia of H. anurae. All measurements are given in μm as means ± standard deviation with range in parentheses. Spores measured using light microscopy as well as organ samples with or without microscopically detectable plasmodia were used for DNA extraction and sequencing of different regions of rDNA. One replicate of H. carassii spores found in C. auratus was used for scanning electron microscopy. For histology, 10 % formalin preserved samples of cyprinid fishes and frog kidneys were processed routinely, embedded in paraffin, sections stained with hematoxylin and eosin (H&E) and examined as specified above.

Electron microscopy
For scanning electron microscopy (SEM), H. carassii spores from an infected C. auratus urinary bladder smear were washed with 0.1 M sodium cacodylate buffer (pH 7.2), collected and fixed with 2.5 % glutaraldehyde in cacodylate buffer. The day of processing, the parasites were washed in cacodylate buffer and centrifuged (800 g, 5 min). Thereafter, the parasites were left to settle onto an ethanol-washed and 0.1 % poly-D-lysine coated coverslips for 30 min and then fixed for 30 min using 2.5 % glutaraldehyde in cacodylate buffer. After rinsing in the same buffer (15 min) the parasites on the coverslip were post-fixed with 1 % osmium tetroxide in 0.1 M sodium cacodylate buffer for 30 min. Coverslips were then washed for 15 min in distilled water, dehydrated in an ascending alcohol series and critical-point dried. Thereafter, the coverslips were mounted on stubs, gold sputtered-coated and examined with a JEOL JSM-7401 F (JEOL Ltd., Japan). For transmission electron microscopy (TEM), selected heavily parasitized paraffin embedded pieces of frog kidneys used previously for histology were first deparaffinized as follows: incubation of the whole paraffin block in 37°C till complete melting of the paraffin, xylene bathtwice for 1 h, 96 % ethanol bathtwice for 1 h, 70 % ethanol bathtwice for 1 h. Tissue samples devoid of paraffin were then post-fixed in freshly prepared 2.5 % glutaraldehyde in 0.1 M phosphate buffer. The specimens were then washed for 1 h in the same buffer, post-fixed in 1 % osmium tetroxide in the same buffer for 3 h and dehydrated in an alcohol series, before embedding in Epon resin (Polybed 812). Sections were cut with diamond knives and stained with uranyl acetate and lead citrate. Observations and imaging were performed using a JEOL 1010 TEM.

rDNA sequencing and cloning
Samples of kidney, ureters and urinary bladders with spores and/or plasmodia of two goldfish, one Prussian carp, five common carp, one Peters' elephantnose fish and one reed frog, H. kivuensis (see Table 1) were stored in 400 μL of TNES (10 mM Tris-HCl (pH 8), 125 mM NaCl, 10 mM EDTA, 0.5 % SDS, 4 M urea) [32]. DNA was digested with 100 μg/ml of proteinase K, overnight at 55°C, and extracted following a simplified phenol-chloroform protocol [1]. The extracted DNA was re-suspended in 100 μL of RNAase/DNAase free water and left to dissolve overnight at 4°C.
Partial SSU rDNA sequences and ITS regions (ITS1, 5.8S and complete ITS2 sequence) were amplified using different primer combinations and PCR conditions (Table 2) [2,[33][34][35][36]. SSU rDNA amplicons were obtained using primers Erib1 + Erib10. If this PCR failed or abundant host tissue was present, a nested PCR with MyxospecF + MyxospecR or MyxGP2F + Act1R was performed. The ITS region was amplified using HofSSUend + ITS-Zschok-Rev. PCRs were conducted in 25 μl reactions with 0.025Uμl −1 Titanium Taq DNA polymerase and 10× buffer which contained 1.5 mM MgCl 2 (BD Biosciences Clontech), with 0.2 mM of each dNTP, 0.5 mM of each primer, and 10-150 ng of template DNA. The PCR cycle conditions consisted of denaturation 95°C 3-5 min, followed by 30-40 cycles of amplification: 94°C for 1 min, specific annealing temperature ( Table 2) for 1-1 min 30 s, and 68°C for 1-2 min, and final extension at 68°C for 8-10 min. After checking for the presence of the expected DNA amplicons in a 1 % agarose gel in sodium acetate buffer, PCR products were purified for sequencing using a Gel/PCR DNA Fragments Extraction Kit (Geneaid Biotech Ltd., USA). Preferably, direct sequencing of PCR products was attempted. Problematic amplicons were cloned into the pDrive Cloning vector (Qiagen PCR Cloning Kit, Germany) and transformed into the competent E. coli strain XL-1. Plasmid DNA was isolated using a High Pure Plasmid Isolation Kit (Roche Applied Science, Germany). PCR products or plasmids were sequenced on an ABI PRISM 3130x1 automatic sequencer (Applied Biosystems, Czech Republic). The overlapping partial corresponding sequences of SSU rDNA and ITS regions were assembled into single contigs in SeqMan II v5.05 (DNASTAR Inc., Madison, Wisconsin).

Specific primer design for PCR-based diagnostics of Hoferellus spp
Based on the Hoferellus SSU rDNA sequences obtained from C. auratus and C. carpio and their alignment with other myxozoan species belonging to the same phylogenetic clade ("freshwater clade" [2]), specific primers (HofcarasF/R and HofK41F/R; Table 2) were designed for a duplex PCR assay. Another PCR assay was designed for the Hoferellus sp. detected in common carp (HofK48F/R). Based on the different Hoferellus ITS region genotypes obtained in C. carpio specific reverse primers (HofK102R and HofK107R, Table 2) were used, combined with the forward primer HofSSUend, in a single-round duplex PCR assay. All specific primers were designed using NCBI\-Primer-Blast (National Center for Biotechnology Information, [37]). No cross-reaction was detected for the new primers designed. Primers were tested for their optimal annealing temperature in a gradient PCR and subsequently applied in its specific PCR assay. Specific PCR assays were conducted as before but in 10 μl reactions with 0.4 U Taq-Purple DNA polymerase and 10× buffer, which contained 1.5 mM MgCl 2 (Top-Bio, Czech Republic) and 72°C for extension. For diagnosis of other Hoferellus genotypes, the same PCR protocol was used but with Titanium Taq DNA polymerase, at 68°C extension temperature. Thereafter, all PCR products were submitted to electrophoresis and positive/negative samples were recorded.

Phylogenetic analyses
Two main alignments were created using MAFFT v6.864 [38] with a L-INS-i strategy and default parameters. Alignments contained newly obtained and published myxozoan sequences retrieved from GenBank. The SSU rDNA alignment contained 2880 characters, whereas ITS region alignment contained 1168 characters. Phylogenetic analyses were performed using maximum likelihood (ML), maximum parsimony (MP) and Bayesian inference (BI). ML was done in RAxML v7.0.3. [39] with the GTR GAMMA model of evolution. MP was performed in PAUP*v4.0b10 [40] with heuristic search with random taxa addition and the TBR swapping algorithm. All characters were treated as unordered, Ts:Tv ratio was set to 1:2 and gaps were treated as missing data. BI was computed in the MrBayes v3.0 [41] with the GTR + Γ + I model of evolution. Posterior probabilities were calculated over 1,000,000 generations via two independent runs of four simultaneous Markov chain Monte Carlo with every 100th tree saved. Tracer v1.4.1 [42] was used to ascertain the length of burn-in

Prevalence of Hoferellus spp
Plasmodia of Hoferellus spp. in the lumina of the kidney tubules were observed with a prevalence of 13.2 % (15/ 114) in C. auratus from two localities and 13.8 % (18/ 131) in C. carpio from five localities, but were not observed in kidney tubules of C. gibelio (0/12). The plasmodia in the kidney tubules were usually smaller than those in ureters and urinary bladder, possessing refractile granules and occasionally forming spores. Larger presporogonic and sporogonic stages, as well as free mature spores were observed in ureters and urinary bladder with a prevalence of 33 % (27/82) in C. auratus from three localities, 14.2 % (15/106) in C. carpio from five localities and 58.3 % (7/12) in C. gibelio from Jihlava (Table 1). These stages were sometimes motile, with large hyaline areas and pseudopodia. In general, microscopic infection prevalence in all goldfish was higher in the ureters/urinary bladders than in kidneys. In common carp, the infection prevalence was similar in kidneys and ureters/urinary bladders, whereas in Prussian carp, only urinary bladders were infected (Table 1). In common carp, mature spores were only found in 2.3 % (3/131) of the fish examined. The earliest spores were detected in the urinary bladder, in February (Třeboň, 1 fish) and in the kidney and urinary bladder in April-May (Chřešťovice and Jindřichův Hradec, 1 fish each). In goldfish, mature spores were detected with a prevalence of 10.5 % (12/ 114, Chřešťovice and Jihlava): 7 kidneys in March-April and June, and 5 urinary bladders in June-August. No obvious signs of disease or kidney enlargement or intracellular stages were observed in any cyprinid fish at any time.
Spores matching the characteristics of Hoferellus located within large plasmodia were observed in the kidney tubules with a prevalence of 10 % (1/10) in G. petersii. The morphology of these spores did not match that of other Hoferellus spp. and hence is described as a new species below.
In wet mounts of kidneys from reed frogs, spores matching original description of Hoferellus anurae Mutschmann, 2004 described from Afrixalus and Hyperolius spp., were observed with a prevalence of 40 % (4/10) and 57 % (4/7) in H. viridiflavus and H. kivuensis, respectively. Fresh mounts and histology revealed the same prevalence, but plasmodia were only discernible in histological sections.

Morphology of Hoferellus spp. from cyprinids
After comparison with previous reports and based on morphological and morphometrical data (see Tables 3  and 4), we ascribed the spores found in C. auratus to H. carassii and the spores found in C. carpio to H. cyprini.

Remarks
Anterior pole of spore end in H. cyprini less pointed than in H. carassii. The irregular posterior spore end, as well as the two posterior small valvular projections are more prominent in H. cyprini than in H. carassii. In both species, spores in different degrees of maturation were observed. Mature spores were more compact and elongate (less round), with clear sporoplasm, fully formed polar capsules and visibly coiled polar filament. Immature spores (Fig. 1e) were more round, less compact and polar capsules were not always in their definitive position and/or the coiled polar filament was not visible.

Hoferellus sp. ex C. carpio
In common carp, a second hoferellid species was detected using molecular tools only (see section "rDNA sequences and phylogenetic results" below) in mixed infection with
Type locality: Nigeria, Africa (not exact location, fish obtained from pet shop).
Material deposited: DNA sample deposited at the protistological collection of the Institute of Parasitology, Biology Centre, Czech Academy of Sciences, České Budějovice, no. IPCAS ProtColl 33.
Sequence data: 2162 bp of SSU rDNA sequence obtained from one infected G. petersii (GenBank Acc. Num. KU141398)

Remarks
Single free spores showed two clear posterior filaments (Fig. 3a). The serrated posterior ends of the spores suggest that caudal filaments might be present, generally indiscernible in light microscopy but potentially indicated in Fig. 3a (right spore). Further material, preferably processed for SEM, is needed to confirm the presence/absence of caudal filaments in this new species.

Redescription of Hoferellus anurae Mutschmann, 2004
Morphology of spores Spores 8 (7.5-8.5) × 7 (6.0-7.5), miter-shaped with pointed anterior pole, longitudinal ridges and usually discernible longitudinal suture. Posterior part of spore possessing two distinct but relatively small projections 0.5-1.2 long, one on each valve, giving the spore somewhat triangular appearance (Fig. 4a). In a negligible number of spores, much smaller projections corresponding with the end of the longitudinal valvular ridges, not clearly discernible posterior filaments, seemed to be present between the two main posterior projections (Fig. 4a). Posterior filaments otherwise indiscernible by light microscopy and considered absent. Some spores more spherical than others (Fig. 4a), possible reflecting degree of maturation. Single binucleate sporoplasm, or two uninucleatenot discernible in light microscopy (nor in TEM preparations). Polar capsules 4.0 × 2.5, pyriform, filling approximately 1/2 of spore (Fig. 4b), coils of polar filament usually oblique to the longitudinal axis of capsule (Fig. 4a).

Localization and morphology of plasmodia
Present only in renal tubules, never observed in glomerular spaces (Fig. 4c). Spore-producing plasmodia very elongate as revealed in longitudinal tubule sections, polysporicup to 19 spores per section plane observed (Fig. 4d).

Intracellular presporogonic stages putatively assigned to H. anurae
Observed regularly in all TEM preparations within epithelial cells of renal tubules, sometimes occupying epithelial cells of tubules containing sporogonic plasmodia in their lumina (Fig. 5a). Although exact cell configuration of these putatively presporogonic stages remained somewhat unclear as serial sections were not analysed, these stages apparently consisted of a single enveloping primary cell, containing several secondary cells (Figs 5b-d).
Epithelial cells of tubules containing only these intracelular stages, but devoid of sporogonic histozoic plasmodia, retained a healthy microvillar layer (Fig. 5c). The intracelular stages were never recognized with certainty in corresponding paraffin sections from which the TEM preparations were made.

Ultrastructure of coelozoic spore-producing plasmodia and myxospores
Apart from mitochondria, various vesicles, spores and electron-dense generative cells, which were often located at the trophozoite periphery, no ultrastructural details could be recognized as a result of sub-standard processing of samples for TEM. Trophozoites filling tubular lumina to various degrees (Fig. 6a), sometimes attached to all surrounding epithelial cells (Fig. 6b), always possessing conspicuous villosities at host-parasite interface (see following section). Spores apparently not formed in pairs, possessing 10-13 longitudinal ridges per valve (Fig. 6d).

Mode of attachment to host epithelium
Trophozoites invariably possessing conspicuous villosities and protrusions wherever they are in contact with microvillar surface of the host tubular epithelium. These represent long, often very fine, protrusions of trophozoite deeply inserted into tubular microvillar zone, as well as invaginations of trophozoite surface (Figs 6e-h). The host-parasite interface is formed by a complex mass of intermingling host microvillar zone and parasite surface of spongious appearance. Microvillar zone of infected tubules seemed pathologically affected by the action of plasmodia in all sections. Although microvillar zone of some tubules was still clearly discernible, the tubules were always irregularly fused and deformed (Figs 6a, e). More often, the tubular zone completely lost its integrity, being composed of irregular protrusions and invaginations intermingled with plasmodial surface folds and projections (Figs 6f-h). In many cases, the close connection between the parasite and the host epithelium was restricted to limited area(s) of plasmodial surface. In such situations, the rest of the plasmodium possessed only little projections and corresponding luminal surface of epithelial cells was rather smooth with only faint remnants of former microvillar zone (Fig 6c). In other cases, plasmodial projections were deeply embedded between altered, apparently swollen tubular epithelial cells (Fig 6b).
Type locality: Nigeria (no exact location provided in original description).

Remarks
Although stated in the original description by Mutschmann [47], posterior spore filaments are not apparent in his micrographs. After thorough examination of numerous fresh mounts and unpublished original photographic documentation from material on which the original description was based (Mutschmann, pers. comm.), we believe the posterior spore filaments, generally distinct  3 and 4). Thus, the presence of only two small projections in H. anurae, is considered herein a species diagnostic feature. Although presence of very fine and short posterior filaments cannot be ruled out based on light microscopical observations, no such structures were detected in our TEM preparations. According to close phylogenetic and biogeographical relationships of vertebrate hosts, we consider our samples conspecific with H. anurae of Mutschmann [47], which is likely a specific parasite of renal tubules of hyperoliid anurans endemic to Sub-Saharan Africa. Thus far, H. anurae is documented from West-Central and East Africa, but is to be expected throughout the distribution range of hyperoliid frogs.

rDNA sequences and phylogenetic results
Partial SSU rDNA sequences obtained in this study are listed in Table 1. The SSU rDNA sequence from C. auratus was identical with the published partial sequence of H. carassii (GenBank Acc. No. JQ801547, also from Czech Republic) and identical with the SSU rDNA sequence from C. gibelio in the present study. All SSU rDNA sequences from Hoferellus spp. in C. carpio were identical except one isolate from a single common carp from Chřešťovice, which showed 2.6 % sequence divergence and will hereafter be addressed as Hoferellus sp. ex C. All the newly obtained SSU rDNA sequences cluster within the freshwater urinary bladder clade (Fig. 7a). The genus Hoferellus is polyphyletic forming two separate The analysis of interspecific SSU sequence distances revealed high minimum sequence dissimilarity (>20 %) between members of Hoferellus s. s. and Hoferellus s. l. (Fig. 8a, y-axis), whereas genetic distances within the Hoferellus s. s. subclade showed 6.6 % maximum interspecific dissimilarity (Fig. 8b, x-axis) in contrast to 26.3 % of maximum interspecific dissimilarity within members of the Hoferellus s. l. subclade (Fig. 8a, x-axis). All Hoferellus spp. revealed similar minimum sequence dissimilarity to other members of freshwater urinary clade (19.8-21.9 %) except of H. gilsoni with the minimum sequence dissimilarity of 15.4 % (Fig. 8a, y-axis). minimum sequence dissimilarity between the three Hoferellus s. s. species (Fig. 8c).
Hoferellus spp. SSU sequences presented inserts in the variable regions. Two inserts of 15 bp and 45 bp respectively in the V2 region were shared by members of Hoferellus s. s. subclade. No inserts were shared by members of Hoferellus s. l. subclade, except for an insert of 25 bp present in H. gnathonemi. Inserts in V4 regions are 27 bp long only in members of the Hoferellus s. s. subclade. M. streisingeri has long inserts in V2, V4 and V7 regions with the longest insert in V4 region (150 bp).
The ITS rDNA sequences of Hoferellus s. s. from cyprinids were obtained by sequencing of clones from fish obtained from different localities ( Table 1): All clones from different goldfish (one site, Jihlava) were identical. Two groups of cloned sequences were observed in carp: 4 clones from Jihlava, 1 from Třeboň, 1 from Chřešťovice and 4 from Jindřichův Hradec (group 1: 0-15.1 % intragroup sequence divergence; H. cyprini) and 4 clones from Chřešťovice and 2 from Nové Hrady (group 2: 1-18 % intragroup sequence divergence; Hoferellus sp.). Sequence divergence between the two groups of clones from carp ranged from 18.2-21.9 %, which was very similar to the variability observed between H. carassii and all carp clones (22.5-24.5 %).
Phylogeny of the cloned sequences of the ITS rDNA region of Hoferellus spp. from common carp and goldfish showed three well-defined branches (Fig. 7b). An individual sample from common carp from Třeboň, K102 (Acc. No. KU141414) provided a single sequence, and was associated with typical H. cyprini spores as described in the present work. All the sequences of clones clustering with this isolate were considered conspecific (Fig. 7b). The other sequences from common carp that did not cluster with this isolate were considered a different species, Hoferellus sp. ex C. carpio (K41 clones Acc. No. Fig. 8 Graphic interpretation of intraspecific and interspecific distances within Hoferellus spp. a SSU rDNA minimum interspecific distances (dissimilarities) between hoferellids and other members of the freshwater urinary bladder clade plotted against maximum interspecific distance within Hoferellus s. s. species and within Hoferellus s. l. species; b SSU rDNA minimum interspecific distances between species of Hoferellus s. s. subclade and species of Hoferellus s. l. subclade plotted against maximum interspecific distance within Hoferellus s. s. species; c ITS minimum interspecific distances among Hoferellus s. s. species plotted against their maximum intraspecific distances KU141406, which were obtained from the same host as the SSU sequence of Hoferellus sp. ex C. carpio, Acc. No. KU141401). Based on the ITS region, intraspecific variability was revealed for H. cyprini and Hoferellus sp. ex C. carpio, but not for H. carassii for which only one sequence was obtained from different clones analysed, however, they all came from the same site. The results from the phylogenetic analyses of the ITS rDNA region provided further support to the SSU rDNA based phylogeny revealing three distinctive separate species in Hoferellus s. s. subclade.

Single and duplex PCR detection of Hoferellus spp. in cyprinids
The results of the combination of PCR assays designed to differentiate between the three species/genotypes of Hoferellus in cyprinids are summarized in Table 1. The assay showed that goldfish and Prussian carp were exclusively infected with H. carassii whereas common carp was infected with H. cyprini and Hoferellus sp. ex C. carpio. Both H. cyprini and Hoferellus sp. ex C. carpio were detected in all localities (except for Bavorov) and higher prevalence was generally observed for H. cyprini (26.7-60 %) than for the other genotype (12.8-50 %). Mixed infections of H. cyprini and Hoferellus sp. ex C. carpio in common carp were detected only molecularly, with prevalences ranging between 13.3-25 % at a range of localities (see Table 1).
PCR detection of H. carassii in C. auratus confirmed the results observed microscopically with higher prevalence in the ureters and urinary bladders than in the kidneys (Table 1). Simultaneous presence of H. carassii in kidney and urinary ducts was detected only in 3 goldfish (2 from Jihlava, in June and 1 from Chřešťovice, in September). H. carassii infections in C. gibelio (June and August 2012) were only detected in ureters and urinary bladders, however, all kidneys were PCR negative. H. cyprini was detected in similar prevalence in kidney and urinary ducts/bladders (40.9 % 9/22 vs 44.8 % 13/29), whereas Hoferellus sp. ex C. carpio molecular prevalence in urinary ducts/bladders (24.2 % 8/33) was higher than in kidneys (12.9 % 8/62).

Discussion
Almost all year round different myxozoan developmental stages were detected in the lumina of the renal tubules of cyprinid fishes in Czech ponds. A high diversity of myxozoan species infecting the urinary system is known from cyprinid fishes, including species of the genera Sphaerospora, Buddenbrockia or Myxidium [48,49]. The lack of spores makes Hoferellus infections difficult to discriminate from other myxozoans, hence, molecular methods are essential for diagnosis, but also to determine true prevalences in case of low infection levels, and to reveal the true diversity of Hoferellus spp. in selected West Paleartic cyprinids.

Hoferellus as a polyphyletic genus
The present study provides 4 new SSU rDNA sequences for members of the genus Hoferellus, including the type species H. cyprini. Phylogenetic analyses of 6 genotypes of nominal species demonstrate that Hoferellus is yet another polyphyletic myxozoan genus. Although most relationships within the freshwater urinary clade, which all Hoferellus spp. belong to, are characterised by low bootstrap support, the polyphyly of hoferellids is confirmed by a strongly supported sister relationship of M. streisingeri with the Hoferellus s. s. subclade. Hence, the Hoferellus morphotype seems to have evolved more than once in the freshwater clade.
High maximum interspecific dissimilarity within members of Hoferellus s. l. is caused due to the divergent SSU rDNA sequence of H. anurae. This species is characterised by a very long-branch in the phylogenetic tree, which could be explained by the longer independent evolution of H. anurae or by the accelerated evolution of this species, potentially due to its occurrence in amphibians rather than fish. Similar long-branch character of SSU is seen in two closely related Ortholinea species. Long branching species can be problematic for reliable phylogenetic reconstruction since their position can be affected by long-branch attraction [50]. Due to the weak support for the position of Hoferellus s. l. their exact position is not determined until other close relatives are found. Hoferellus s. l. cluster closely with several marine Ortholinea species, which would suggest several evolutionary switches between marine and freshwater environments. The most parsimonious explanation of this phylogeny would be a more basal position of Hoferellus s. l. and a closer relationship with marine Ortholinea spp.

Host specificity of Hoferellus s. s. from West Palearctic cyprinids
Our molecular identification and PCR screening results of Hoferellus spp. in cyprinid fishes of Czech ponds clarifies the much-discussed identity of these species [12,13,17] (see Tables 3 and 4). The type species, H. cyprini, and the second genotype detected, appeared to be specific for common carp, C. carpio, and they did not infect goldfish or Prussian carp. H. cyprini is slightly more prevalent than Hoferellus sp. ex C. carpio, although mixed infections are common. Despite the large sample size, we observed spores of only a single Hoferellus species in common carp. Sequences from these always included the genotype here identified with H. cyprini. Only by molecular tools was it possible to detect both species in carp, revealing that Hoferellus sp. ex C. carpio more frequently infects the ureters and urinary bladders, whereas H. cyprini was detected all along the excretory system, suggesting that these taxa show sympatric speciation by exploiting slightly different host habitats, and they likely speciated relatively recently.
A different scenario was observed in H. carassii. This species was originally described from Prussian carp by Akhmerov [22] and reported from different hosts according to morphological identification (see Table 3). We confirmed by species-specific duplex PCR that H. carassii infects at least two different host species of the genus Carassius, C. auratus and C. gibelio, but it does not infect common carp, C. carpio. A relaxed host-specificity for the two closely-related Carassius species [51] that were considered subspecies before, was also observed in other, generally highly host-specific genera, such as Sphaerospora [48]. Furthermore, SPF (specific pathogen free) common carp exposed to the actinosporean stage of H. carassii, produced by the invertebrate host did not result in infection under experimental conditions [25]. H. carassii seems to develop differently in the two Carassius hosts: Previously, H. carassii in C. auratus was found in kidney tubules and urinary ducts, whereas in C. gibelio it infects only the urinary bladder [19]. This agrees with our results of both microscopy and PCR in C. gibelio. This different infection pattern may indicate hostdependent variability in habitat selection of H. carassii, potentially reflecting the onset of a speciation process.
Pre-sporogonic development of Hoferellus s. s. and the identity of enigmatic intracellular stages An annual life cycle was assumed for H. cyprini in common carp, with early intracellular stages in the tubular epithelium during summer, mature intracellular stages in the syncytium containing up to quaternary cells in autumn and winter, that are released gradually into the lumen from winter to spring, producing mature spores in spring [19,27]. Similarly, an annual cycle was described in H. carassii from goldfish: intracellular stages in October-December, luminal stages in January-February and spores in April [17,20]. In the present study, common carp and goldfish were sampled from February to November. Although abundant coelozoic (luminal) stages were found for both species throughout the sampling period, no intracellular stages were detected at any time or locality in the present study. A possible explanation therefore may be a low infection intensity in the fish analysed in the present study, potentially leading to a low number and thus great difficulty in detecting these stages. Prevalence data for Hoferellus intracellular stages are rare but existing ones are relatively low: 1/12 common carp, January [52], 1/42 gibel carp, October and 3/12 goldfish, April-November [19]. Intracellular myxozoan stages in renal tubules epithelia in carp and coho salmon were suggested to be abortive extrasporogonic developmental stages of other myxozoan species, e.g. Sphaerospora or Myxidium, which stages seemed to degrade, instead of following through to sporogony [16,29]. In a similar manner, Hoferellus sp. ex C. carpio may be unable to complete its development and produce spores, as they were never detected microscopically. It may well be a related species that sporulate in a different, likely related host, and hence only initial stages of development can be found in carp. For example, Myxobilatus gasterostei (Parisi, 1912), a species with intraepithelial development in the kidney tubules of sticklebacks, was reported in the blood of C. carpio and C. auratus as an alien species that does not form spores in these hosts [6,48]. Alternatively, intracellular stages might also belong to a transient developmental stage of the malacosporean Buddenbrockia sp. 2, a common species in the kidney of common carp and goldfish [48,49,53]. In the present study, intracellular presporogonic stages with several secondary cells were described in frog kidneys. Unfortunately, no molecular proof could be provided in this study and they were putatively assigned to H. anurae, a member of Hoferellus s. l. Further molecular studies are needed to identify the species to which the enigmatic intracellular stages belong to.

Pathogenicity of nominal Hoferellus spp
H. carassii is suggested to be the causative agent of KED in goldfish [19,20]. The renal damage related to KED is caused by H. carassii stages invading the epithelial cells of the renal tubules, with disease symptoms only observed in pet fish populations kept in small ponds, but not in natural fish populations [19]. Despite the high prevalence of infection in goldfish kept in small ponds for sale as ornamental fish, no mortalities or signs of disease were observed in the present study or in a previous study [25]. A possible explanation could be the mild infections in the goldfish from our study, which also disabled the detection of the intracellular stages.
In 2004, Mutschmann described 'frog kidney enlargement disease' in hyperoliid anurans from the pet trade and identified its causative agent as H. anurae, which he formally described and named in the same study. Although we did not detect any gross renal pathology even in heavily infected (but otherwise obviously healthy) frogs from a wild population, we detected pathological changes on an ultrastructural level. The pronounced deterioration of the microvillar zone and swelling of epithelial cells of infected renal tubules might imply that H. anurae has a potential to cause disease under certain conditions, such as host immunosuppression or heavy infections. This might explain the apparent absence of the disease in wild frog populations in contrast to the severe impact of the infection on frogs likely subjected to suboptimal and stressful conditions associated with the pet trade.

Hoferellus s. l. diversity and host-parasite interactions
Morphologically, members of Hoferellus s. l. have slightly different spore shapes. H. gilsoni and H. gnathonemi parasitize teleost fishes and have round or subspherical spores produced in pairs, whereas H. anurae infect frogs, possess pyramidal or miter-like spores which are flat at the posterior end and do not develop in pairs. H. gilsoni and H. anurae seem to share a rather unique mode of attachment to the epithelia of its host, the urinary bladder [30] and kidney tubules (present study), respectively.
Phylogenetically, all three species, H. gilsoni, H. anurae and H. gnathonemi clearly clustered separately from Hoferellus s. s. However, the phylogenetic relationship between them was not resolved with the presently available sequence data: our SSU rDNA phylogenetic results showed no clear clustering of the three species. Therefore, we suggest their temporal classification as Hoferellus s. l. before more morphological and molecular information is available for substantiated taxonomical rearrangements.
Taxonomy and species currently assigned to genus Hoferellus Berg, 1898 The genus Hoferellus was originally described by Doflein in 1898 [43] as Hoferia, with Hoferia cyprini as type species. However, the name Hoferia, had been assigned to an extinct genus of molluscs by Bittner in 1894 [54]. The genus was then renamed as Hoferellus (nov. nom. pro.) by Berg in 1898 [8]. The genus Mitraspora Fujita, 1912 is a further synonym of Hoferellus according to Lom 1986 [12]. The establishment of Mitraspora was probably based on a different suture position on the spores [13]. In the original species description of H. cyprini and H. carassii the suture was wrongly described as in the same plane as that of the polar capsules. Mitraspora was established with Mitraspora cyprini (Fujita, 1912) as type species, to include species that had resemblance with Hoferellus species but have a suture perpendicular to the plane of the polar capsules. Mitraspora was synonymized with Sphaerospora by Shulman (1966) [10] probably because in sphaerosporids the suture position is also perpendicular to the polar capsules. Mitraspora was considered again as a separate genus by Lom & Noble (1984) [11], assigning H. carassii as type species of the genus Hoferellus and M. cyprini as type species of Mitraspora. Finally, the genus Mitraspora was considered a junior synonym of Hoferellus [12,13,27] with H. cyprini as the type species.
All Hoferellus spp. (syn. Mitraspora) described to date and for which DNA sequence data is missing, should be addressed as Hoferellus incertae sedis until SSU rDNA sequences are obtained and they can be ascribed to Hoferellus s. s. or s. l. H. anurae, H. gilsoni and H. gnathonemi will remain Hoferellus s. l., although they may resolve in new genera, at a later stage.
According to the review by Lom