Attachment and early development
The present results regarding the early development of Buddenbrockia worms confirm most earlier observations made on the basis of light and electron microscopy [2, 22–25]. The distribution of attached stages inside the host supports the view that the soft, exposed parts of the host body wall (for example, the tentacle sheath) and the gut are the main entry portal for the myxozoan spores, rather than the cystid wall, which is covered by a gelatinous or chitinous ectocyst. The presence of parasite stages (1) anchored in the ECM (2) attached to the surface of the coelomic epithelium and (3) free-floating in the body cavity suggests that very early and unicellular stages are able to migrate through the coelomic epithelium. Whether these develop into attached or free-floating stages may depend on their position within the host when growth commences. Earlier studies suggested that the onset of worm growth may be triggered by warmer temperature or availability of nutrients, since worms were observed to develop when subject to these conditions in a laboratory mesocosm . There is strong evidence that such conditions trigger development of the closely related malacosporean Tetracapsuloides bryosalmonae[29–31].
The host tissue reaction may additionally strengthen the attachment of the worm, as it involves thickening and reinforcement of the cytoskeleton of the peritoneal cells that surround the attachment area. As worms of all lengths were found attached as well as free-floating, we suggest that detachment is not a scheduled event during development but occurs haphazardly, possibly facilitated by the increasingly vigorous movements of the worms. This hypothesis is supported by a higher proportion of detached older worms. The undifferentiated state and slightly ruptured appearance of the tissues at the proximal tip of detached worms renders it possible that a few cells could remain in the “wound” after detachment. Such material may therefore result in ongoing covert infection and the future growth of another worm. Note that covert infections are also likely to be achieved by the persistence of early stages, which we have observed in host tissues simultaneously with developing and mature worms. Such covert infections may promote long-term persistence in hosts such as occurs in T. bryosalmonae[31, 32].
We have not found any evidence for the idea that the worms’ muscle cells could be derived from bryozoan muscle cells, as suggested by Morris and Adams . The latter look rather different, and, in addition, muscle development occurs in the distal end of the worm. Furthermore, bryozoan nuclei are considerably larger than those of Buddenbrockia (see also  for ultrastructural differences), making potential chimeras easily detectable. We thus conclude that the musculature is a native Buddenbrockia feature that is likely to have been retained from a free-living cnidarian ancestor. In addition, our data do not support the idea that early multicellular stages form by the accumulation of migratory unicellular stages rather than by mitosis . This would require the unicellular stages to move horizontally throughout the ECM, and a progression from loose clusters of cells to densely packed tissuelike stages would be expected. Such a progression is not evident in our data.
The present data demonstrate that the connecting cells persist in the mature worm and do not contribute to sporogony, as also shown by ultrastructure . The possibility that they serve a mechanical function is supported by their attachment to the surrounding ECM, their stretched appearance and their pronounced actin cytoskeleton. However, the possibility of a neuronal function, as suggested by Schröder , cannot be ruled out completely.
The presence of a polarised primary body axis
Our study shows that, during ontogeny, Buddenbrockia worms acquire a distinct polarity along their primary body axis, which is reflected by directional growth and a gradient of tissue differentiation. The inner cells as well as the musculature at the distal tip of developing worms are more differentiated than at the proximal tip. This polarity persists in later stages, and, in some fully mature worms, a porelike opening appears at the distal tip. However, from the present data, it cannot be determined whether this is a real pore or an opening as a result of rupture. Although internal cells at the distal pole sometimes differ in their morphology, we found no indication that these represent spermatids . Anti-phospho-histone H3 staining does not give evidence of the presence of a distinct growth zone at either end of the worm. However, as the proportion of nuclei labelled with this method is low in comparison to methods such as bromodeoxyuridine labelling, which integrate over longer time periods, the latter might provide a more detailed picture in future studies.
Because Buddenbrockia lacks a gastrulalike stage, an intestinal tract and evidence of nervous structures, it is difficult to relate this axial polarity to that of other animals. Recent experimental and gene expression data (especially those based on Wnt/β-catenin signalling) demonstrate that the development of polyps from planulae and of asexual buds is in accord with early morphology-based hypotheses for a homology of the cnidarian oral-aboral axis and the bilaterian anteroposterior axis [33–36]. These data also suggest, although with lower confidence, that the cnidarian oral pole may correlate with the bilaterian posterior pole. Together with findings that growth at the posterior pole is most likely an ancestral character in bilaterian animals , this implies that the proximal tip of Buddenbrockia may correspond to the bilaterian posterior pole and thus the cnidarian oral pole. However, in cnidarians, no generalised pattern of either anterior or posterior growth has so far been demonstrated.
Another question is when and how polarity is determined. In cnidarians and bilaterians, the main body axis is either identical to or oriented at a particular angle to the embryonic animal-vegetal axis. This is achieved, for example, by gradients of maternally expressed transcription factors in the egg or the position of the egg with respect to the ovarial tissue (reviewed, for example, in [38, 39]). However, the unicellular stages found in the bryozoan tissue are most likely derived from sporoplasms that came from spores produced in the vertebrate intermediate host . They are amoeboid and do not show any recognisable polarity before mitotic divisions leading to early multicellular stages. It is therefore likely that polarisation is induced by external factors. This might be achieved via the orientation of the early stages in the bryozoan host tissue, such as with growth directed away from the basal lamina. Such a scenario could be tested by detailed comparison with closely related, saclike malacosporeans whose trophic stages lack a distinct body axis, such as the Buddenbrockia parasite of Cristatella mucedo; Buddenbrockia allmani in Lophopus crystallinus; or Tetracapsuloides bryosalmonae, which predominantly parasitises Fredericella sultana. For the latter species, data on early development indicate that presaccular stages are situated on the surface of the peritoneum rather than underneath as in Buddenbrockia, rendering the previous explanation possible.
Mesodermal musculature and germ layers
As unequivocally indicated by the presence of nuclei in the muscle cells, the musculature of the Buddenbrockia worm is formed by independent myocytes and not by epitheliomuscular cells as in most other cnidarians [41, 42]. If the epidermis is regarded as ectodermal and the inner epithelium (although functioning only as reproductive rather than digestive tissue) as endodermal, then the muscle cells are, at least by topological definition , mesodermal as they reside in the ECM between the two epithelial layers and are not connected to the latter. We note that the same argument applies if adaptation to parasitism has involved an inversion of ectoderm and endoderm as occurs in Polypodium hydriforme. This may have bearing on questions concerning the evolution of metazoan body plans, since the diploblastic vs. triploblastic organisation of the last common ancestor of Cnidaria and Bilateria is highly debated (see, for example, [19–21, 45]). Bilaterian mesoderm is usually characterised by (1) topology, (2) germ layer-specific derivation during gastrulation or (3) the presence of a common set of regulatory genes . The commonly cited example for topological mesoderm in Cnidaria is the entocodon, a tissue that invaginates during hydrozoan medusa bud formation and gives rise to an independent striated subumbrellar musculature in the medusa (summarised in ). The musculature in free-living stages of the parasitic cnidarian Polypodium hydriforme also appears to be topologically mesodermal [46, 47].
In contrast to the hydrozoan entocodon, the mesoderm of Buddenbrockia is formed early in ontogeny and might thus qualify as a bona fide germ layer potentially homologous to mesoderm in bilaterians. It also develops closer to the inner endodermal layer than to the outer epidermal layer. However, the lack of a distinct cleavage pattern, blastula-like stage and gastrulation-like process hinders further comparisons with cnidarian or bilaterian development. Of course, a mesodermal as opposed to a myoepithelial musculature may have functional advantages in a wormlike organism (see for example, ), thus promoting a convergent origin of this character in Buddenbrockia and bilaterians. The vermiform parasitic sea anemone Edwardsiella lineata, however, has typical anthozoan endodermal longitudinal muscles . Gene expression data from Buddenbrockia should enable further insights into, for example, tissue homologies and axial patterning.
Tetraradial symmetry and chirality
The arrangement of the muscle cells within the muscle blocks demonstrates that Buddenbrockia worms are characterised by tetraradial symmetry. Although traditionally regarded as radially symmetric, recent evidence, especially from developmental and gene expression studies, suggests that the ancestor of all cnidarians was a bilaterally symmetrical animal, a pattern still reflected in the organisation of many recent anthozoans [50–52] (but see ). Tetraradial symmetry must be regarded as having evolved in the Medusozoa, which form the sistergroup to Anthozoa . The corroboration of a tetraradial symmetry in Buddenbrockia worms therefore provides further support for a medusozoan affinity of myxozoans .
A further interesting finding is the presence of a consistent handedness or chirality (mirror asymmetry) in the arrangement of the muscle fibres, which align in a right-handed thread in all individuals examined in this study. The coincidence of radial symmetry, or mathematically more precisely, rotational symmetry, together with mirror symmetry, may jointly characterise objects, with examples being a regular geometric star or the majority of radially symmetrical animals. However, the two types of symmetry are not necessarily linked. Many rotationally symmetrical objects are chiral and exhibit rotationally repeated elements with no mirror symmetry. Such chiral forms exist in two (dextral vs. sinistral) enantiomorphs (see  for a review of symmetry patterns).
Chirality is a well known phenomenon in many bilaterian animals (e.g. molluscs, annelids, pterobranchs, nematodes, vertebrates) , but has so far only rarely been described in non-bilaterians. An interesting non-bilaterian case includes certain conulariids, which are a fossil group inferred to have been scyphozoans. Thus, the tetraradially symmetrical skeleton of the conulariid Metaconularia anomala shows torsion along the longitudinal axis . M. anomala also exhibits a strong preference for a sinistral coil . Other nonbilaterian examples include the siphonophore Bargmannia elongata, in which bud formation leads to consistently asymmetrical colony forms , and fossils probably belonging to the ctenophoran lineage and which possess arms that coil preferentially dextrally .
Fixed chiralities such as the one described here in Buddenbrockia are in almost all cases heritable . So, is it likely that chirality is adaptive? A similar angle of the muscle fibres throughout the body is probably strongly selected for because of its functional significance for spiralling movements. However, such movement should be achieved regardless of whether the angle is dextral or sinistral. A possible explanation for the dominance of one chiral form might be if concerted movements of multiple worms within a host minimise interference.
How is the chirality established in Buddenbrockia worms? In most bilaterians, where the developmental mechanism leading to chirality is known, symmetry breaking occurs after the establishment of the dorsoventral axis (see, for example, ). However, a deeper underlying mechanism that establishes chirality on a subcellular level is generally inferred. This has not been clearly identified and may differ from case to case. The main theories are that cellular chirality is a result of (1) molecular chirality of the cilium, (2) cytoskeletal asymmetries leading to a differential distribution of ion channels and/or pumps on one side of a blastomere or (3) nonidentical blastomeres produced by different epigenetically imprinted patterns . The ciliary model can clearly be ruled out for Buddenbrockia, as cilia and centrosomes are lacking in all myxozoans . As the chirality is reflected only in the arrangement of the muscle cells, it could be explained by a directional shift in the cleavage plane in mitoses leading to the muscle cell lineage.