Systematic comparison and reconstruction of sea urchin (Echinoidea) internal anatomy: a novel approach using magnetic resonance imaging

Background Traditional comparative morphological analyses and subsequent three-dimensional reconstructions suffer from a number of drawbacks. This is particularly evident in the case of soft tissue studies that are technically demanding, time-consuming, and often prone to produce artefacts. These problems can partly be overcome by employing non-invasive, destruction-free imaging techniques, in particular micro-computed tomography or magnetic resonance imaging. Results Here, we employed high-field magnetic resonance imaging techniques to gather numerous data from members of a major marine invertebrate taxon, the sea urchins (Echinoidea). For this model study, 13 of the 14 currently recognized high-ranking subtaxa (orders) of this group of animals were analyzed. Based on the acquired datasets, interactive three-dimensional models were assembled. Our analyses reveal that selected soft tissue characters can even be used for phylogenetic inferences in sea urchins, as exemplified by differences in the size and shape of the gastric caecum found in the Irregularia. Conclusion The main focus of our investigation was to explore the possibility to systematically visualize the internal anatomy of echinoids obtained from various museum collections. We show that, in contrast to classical preparative procedures, magnetic resonance imaging can give rapid, destruction-free access to morphological data from numerous specimens, thus extending the range of techniques available for comparative studies of invertebrate morphology.


Background
For centuries, comparative zoomorphological analyses have formed the backbone of phylogenetic inferences. Data on the external and internal morphology of specimens were obtained by dissecting and sectioning selected representative species. These traditional procedures are technically demanding as well as time-consuming and irretrievably alter or even destroy the specimen. Further complications arise from the difficult three-dimensional (3D) reconstruction of the structures observed. Artefacts, for example deformation or compression of histological slices, and the problematic alignment of sections make 3D reconstructions a laborious and often highly subjective procedure. The problems encountered are further aggravated by the logistic challenges that arise when comparative studies on selected species of a whole taxon are to be carried out. For example, in case of sea urchins (Echinoidea, Echinodermata), some species are deep-sea dwellers or confined to remote regions of the world. It is therefore a considerable problem to obtain these animals in a freshly fixed state. Understandably, the desirable use of specimens from museum collections is usually not permitted for invasive and destructive analyses.
The application of non-invasive imaging techniques such as micro-computed tomography (µCT; see, for example, [1][2][3][4][5][6][7][8]) or magnetic resonance imaging (MRI; see below) has substantially increased in basic morphological research. Whereas µCT is mainly used to display hard tissue such as bone or calcite structures, MRI provides images depicting primarily soft tissue anatomy. MRI is based on the principles of nuclear magnetic resonance (NMR) and is predominantly used to measure the distribution of hydrogen protons within a sample. Image contrast is achieved due to the different physical properties of tissues. Both µCT and MRI produce standardized digital data, principally permitting an interactive access through virtual collections or voxel libraries as envisaged previously [9,10]. This data represents the unaltered structural composition of the specimen, in contrast to classical histological or dissection techniques and apart from fixation. In the light of these and other recent technological advances, Budd and Olsson [11] refer to the renaissance of morphology, calling for a more extensive use of non-invasive and non-destructive imaging techniques and 3D applications. Since many of the specimens in natural history collections are conserved in alcohol, forming the so-called wet collections, non-invasive imaging techniques might be employed to reveal the internal and external anatomy of preserved, scientifically irreplaceable specimens. This extended use of museum specimens for comparative anatomical studies would allow for a substantial increase in taxon sampling on one hand, and on the other would minimize the ecological impact of such analyses.

Ziegler and Angenstein [31] used living and freshly fixed specimens of the sea urchin
Psammechinus miliaris (Müller, 1771) and showed that members of this taxon are likely to be suitable for detailed MRI studies. These studies yielded imaging results comparable to manual dissection and were correlated with histological data. To achieve better resolved images of the complex internal anatomy of sea urchins, the application of higher magnetic field strengths was proposed [32]. However, all of the studies mentioned above were restricted to the analysis of single specimens or species in a non-systematic context.
For systematic analyses, however, sea urchins appear particularly suitable. Like sea stars or sea cucumbers, they are part of a major deuterostome taxon, the Echinodermata, which shares a common ancestor with diverse groups such as acorn worms (Enteropneusta), tunicates (Tunicata), and vertebrates (Vertebrata) [33]. Owing to the calcified skeleton found in sea 5 urchins, palaeontologists have access to a rich fossil record permitting thorough groundtruthing of phylogenetic hypotheses that are based upon molecular or morphological datasets [34,35]. Recently, the importance of these animals for evolutionary and developmental inferences has been further emphasized by results obtained through the sequencing of the genome of the purple sea urchin, Strongylocentrotus purpuratus (Stimpson, 1857) [36].
Sea urchins are spherical, oval or flattened free-moving echinoderms covered with spines.
Their soft tissue anatomy is characterized by a dominant digestive tract, a varying number of gonads, and the highly specialized water vascular system with its external appendages, the tube or ambulacral feet. The space between the major organ systems is composed of the main body cavities (the so-called perivisceral coeloms) that are filled with coelomic fluid [37].
Therefore, echinoids possess several of the internal structures encountered in most other taxa, although a number of highly specialized organs are present as well. This taxon can thus serve as a model for analyses using non-invasive imaging techniques on a broad systematic scale.
The purpose of our study was to explore the possibility of generating high-resolution 3D MRI datasets from selected specimens of a major invertebrate taxon. For the first time, a noninvasive imaging technique is employed for systematic comparative analyses. We show that MRI permits the rapid acquisition of digitized morphological data on a broad scale and the construction of 3D models that can be made interactively accessible on the Web. In addition, our approach can be employed to compare the internal anatomy of sea urchins to evaluate such datasets for evolutionary inferences.

Establishment of imaging conditions
Initially, freshly fixed as well as museum specimens of Psammechinus miliaris (Müller, 1771), a common species in the North Sea, were used in order to establish imaging and contrasting protocols for broader systematic analyses. Living specimens of P. miliaris were made available through the Biologische Anstalt Helgoland, Germany (BAH). Several dozen sea urchins were kept in the aquarium facilities of the Institut für Biologie, Freie Universität Berlin, Germany, for subsequent scanning and contrasting experiments.
The first set of analyses was conducted to compare different specimens of freshly fixed sea urchins. Imaging was performed initially at a resolution of (81 µm) 3 using different imaging protocols with specimens of almost identical size. This analysis showed that although slight differences in anatomical makeup did occur, localization, arrangement and overall shape of the internal organs did not differ significantly. In the specimens depicted ( Figure 1A and B), this is exemplified by the almost identical shapes of both Aristotle's lanterns and the localization and shape of the digestive tract components. The structure of the wall of the sea urchin lower gut loop (also termed the stomach) was undulated in both specimens, whereas the horizontal sections of the upper gut loop (also termed the intestine) showed a smooth structure. The digestive tract of one specimen was partly filled ( Figure 1A), while the digestive tract of the other specimen was empty, apart from tiny objects ( Figure 1B). Another striking similarity was the almost exactly identical location of the inner and outer marginal sinus of the haemal system in both specimens. However, some differences were discernable, for example the distinct shapes of the ampullae and the thickness of the interpyramidal muscles, indicating that, ideally, several specimens of a given species should be scanned.
Similar comparative experiments were performed on a total of 18 echinoid species (data not shown). In all cases, the overall internal architecture of two specimens of a given species resembled each other more closely than that of other species. The next set of experiments served to assess the effects of a gadolinium-based contrast agent (Magnevist, a standard in clinical and pre-clinical MRI studies) on image quality. Using the same specimen of P. miliaris and the same imaging protocol, we found that the application of the contrast agent resulted in enhanced contrast, although some thin-walled structures, such as the ampullae or the oesophagus, became less visible ( Figure 1C and D). Another advantage of the use of this contrast agent was that some of the artefacts caused by paramagnetic gut content were reduced due to the stronger water signal. Therefore, most specimens were scanned both before and after the application of the contrast agent (see Additional file 1 for details).
The final set of experiments for establishing a standard set of imaging conditions was designed to compare imaging results obtained from freshly fixed and museum specimens of the same species. Prior to the present MRI studies, several museum specimens from the wet collection of the Systematische Zoologie am Museum für Naturkunde, Berlin, Germany (ZMB) were dissected in order to assess the state of preservation after decades or even more than a century of alcohol conservation (Strongylocentrotus dröbachiensis (Müller, 1776), Sphaerechinus granularis (Lamarck, 1816), Schizaster lacunosus (Linnaeus, 1758), and Brissopsis lyrifera (Forbes, 1841)). Their state of preservation was surprisingly good and seemed mainly to depend on the mode of collection, the fixation mode, and the subsequent preservation in alcohol. Following the dissections, MRI experiments using freshly fixed and museum specimens of P. miliaris were carried out. Structures visible in the freshly fixed specimen ( Figure 1E) were recognizable also in the 135-year-old museum specimen ( Figure 1F). The size and location of the anatomical features varied slightly, but this had been observed among different specimens of freshly fixed animals as well. Despite these differences, the results confirmed our assumption that museum material, even if more than a hundred years old, can be used for systematic MRI studies.

Scanning of selected members of high-ranking echinoid subtaxa
After an initial assessment of the suitability (that is, size, state of preservation, and availability) of specimens in natural history collections worldwide, MRI studies were carried out using selected species of the 14 currently recognized echinoid subtaxa (orders). In three cases, freshly fixed animals were employed for the analysis (see Additional file 2 for details).
The full datasets from two 'regular' sea urchins (Eucidaris metularia (Lamarck, 1816) and P. miliaris) as well as two irregular sea urchins (Echinoneus cyclostomus (Leske, 1778) and Echinocyamus pusillus (Müller, 1776)) are presented here as interactive videos (Additional files 3-6; Figure 2 shows images of selected sections). The four image datasets range in spatial resolution from 20 × 18 × 18 µm 3 to (86 µm) 3 and represent the current state-of-the-art in high-field MRI at a field of view of up to (3.3 cm) 3 . In E. metularia ( Figure 2A) and P. miliaris ( Figure 2B), the pentameric symmetry of echinoderms becomes easily visible when looking at the five gonads or the lantern muscles that form a five-tipped star. Moving down from the aboral to the oral side, the following internal and external structures can be identified, amongst others (Additional files 3-6): madreporic plate, gonads, spines, intestine, compass elevator muscles, Aristotle's lantern with corresponding muscles, and the stomach.
Some parts of the gut of P. miliaris were filled with paramagnetic sediment of unknown composition (possibly iron or manganese residues) that caused artefacts by altering the homogeneity of the magnetic field inside the MRI instrument.
In E. cyclostomus ( Figure 2C) and E. pusillus ( Figure 2D), the secondarily developed bilateral symmetry found in the Irregularia is obvious when looking at the gonads (reduced to four).
The digestive tract of E. cyclostomus is characterized by a dominant gastric caecum, localized at the beginning of the stomach, and a long rectum, leading down to the anus which is situated at the oral side of the animal. E. pusillus, in common with all members of the Clypeasteroida, the so-called sand dollars, possesses a modified Aristotle's lantern. Other structures that can be identified in the irregular species include relatively small spines, one dominant (E. pusillus) or two huge (E. cyclostomus) gut loops, filled with debris and detritus, and an axial complex running straight down from the madreporic plate to the adoral part of the oesophagus (E. cyclostomus). In addition to soft tissues, the delineation of calcareous structures caused by the signal-providing surrounding fluids and tissues adds some information on hard-part anatomy to the virtual sections as well.
Although several specimens of the echinoid taxon missing in our study, the Echinothurioida (or leather urchins), were found in museum collections, their shape prevented successful soft tissue imaging. These sea urchins possess a flexible calcite endoskeleton as well as specialized internal muscles that contract when the animal is fixed, resulting in a pancake-like shape of the organism. For a reliable comparison of internal structures it would be necessary to scan specimens prior to contraction of these powerful muscles.

3D reconstruction of selected internal organs
The 3D reconstructions presented here are limited to the major soft tissue structures identifiable in all datasets, if present in the respective species ( Figure  MRI also revealed a striking difference regarding the architecture of the gastric caecum. Except for the two clypeasteroid subtaxa Laganina and Scutellina, this structure can be found in all irregular taxa ( Figure 4C-E), located at the beginning of the stomach. Its presence correlates with the feeding habits of this group of sea urchins: most of these animals are partially or completely covered by sediment which they ingest to filter out organic material.
Although the precise function of the gastric caecum is still a matter of discussion, it appears to have a major role in digestion (see, for example, [38,39]). In specimens of comparable size, substantial differences in its size and shape were found ( Figure 5). The highly reduced gastric caecum consisting of numerous smaller blindly ending sacs that can be observed in Echinolampas depressa Gray, 1851 ( Figure 5B) appears to be a common characteristic which sets the Cassiduloida apart from other taxa ( Figure 5A, C and D) of the Irregularia. This is emphasized by the presence of a similar, highly reduced gastric caecum in the cassiduloid species Cassidulus caribearum Lamarck, 1801 (data not shown; as described also by Gladfelter [40]) and Rhyncholampas pacificus Agassiz, 1863 (as shown in [41]). A comprehensive description and comparative analysis of sea urchin internal anatomy will be published elsewhere (Ziegler et al., publication in preparation).

Interactive viewing of selected species in three dimensions
To illustrate the opportunities offered by digitized reconstructions, Figure 6A-C shows different views of the 3D model of a member of the Cidaroida (or pencil urchins), E. metularia. This widely distributed Indo-Pacific species possesses a moderately thick calcite endoskeleton, large primary and short secondary spines. Its internal organization is characterized by the presence of five bushy Stewart's organs, five stalked gonads and a short intestine. The drawing presented in Figure 6D was made by Stewart [42] from a closely related species, Cidaris cidaris (= Dorocidaris papillata) (Linnaeus, 1758), where he depicts the newly discovered organs that were later named after him. A comparison of his drawing with the digital model in Figure 6C

Methodology
The initial phase of our study served to determine the proper imaging conditions using highfield MRI instruments. In order to achieve resolutions below (100 µm) 3 , we employed 7 T and 17.6 T small animal scanners. We were able to obtain resolutions ranging from 20 × 18 × 18 µm 3 to (86 µm) 3 with specimens ranging from 5 mm to 3.4 cm in size. Therefore, in terms of resolution, our data is comparable to results derived from manual dissection techniques in combination with stereomicroscopic observation. Susceptibility artefacts were primarily seen in the sediment-feeding irregular species, with the gut content being the source of these artefacts (see, for example, Additional file 6, Echinocyamus pusillus (Müller, 1776)). What sediment component is causing which type of artefact needs to be determined in mineralogical studies on freshly fixed specimens. The effects of the contrast agent made it possible to reduce the negative effects of these artefacts in some cases ( Figure 1C  µCT techniques that are based on X-rays, the living specimen is not harmed during the scanning process. Extensive morphological or developmental studies can thus be carried out, although anaesthetic substances may have to be applied to reduce movement artefacts during the long scanning times that are needed for high-resolution datasets and 3D reconstructions. Another very important advantage of non-invasive imaging techniques in general is the fact that the digitally acquired image stacks do not have to be manually aligned, as is the case with 13 digital pictures from histological or electronmicroscopical sections. Although these datasets can be used to generate highly resolved digitized 3D models of invertebrates as well (see, for example, [47,48]), these analyses are extremely demanding in terms of specimen preparation, slice alignment and 3D reconstruction. In comparison, 3D reconstruction of MRI data is simple and highly accurate, but the segmentation process carried out for the present work is still subjective and our primary hypotheses regarding organ designation had to be evaluated experiments on other invertebrate taxa should be performed can only be partially derived from our study, since imaging results of fixed specimens depend on several factors, some of them still not fully understood (for example, susceptibility artefacts or changes to organ systems after fixation). Another drawback is the fact that only limited in vitro staining possibilities are available for MRI studies, although some substances (for example manganese) can be used for in vivo staining [28]. A further disadvantage of MRI is that specimens cannot currently be scanned in alcohol, since this would require spectroscopic imaging techniques resulting in much longer scanning times at the desirable resolutions below (100 µm) 3 . In addition, in the case of sea urchins, the 3D models generated (see, for example, Figure 6) are practically useless for taxonomic purposes since sea urchin taxonomy is currently based entirely on hard parts. Resolutions well below (10 µm) 3  On the other hand, the access to MRI instruments does not appear to be a limiting factor.
There are currently more than 1000 experimental small animal MRI instruments worldwide available for pre-clinical imaging. Since all high-field MRI scanners consist of superconducting electromagnets, overnight and weekend scans as performed during this study would permit an optimal use of this resource.

Internal morphology of sea urchins
Despite the existence of extensive fossil, morphological and molecular datasets for this taxon, comparatively little systematic knowledge has been gathered on the internal anatomy of sea urchins. However, some authors have greatly extended our knowledge about the internal anatomy of selected, readily available sea urchin species such as Cidaris cidaris (Linnaeus,  [57] or Echinocardium cordatum (Pennant, 1777) [38], not to mention the extensive works of Alexander E Agassiz, Hubert L Clark and Theodor J Mortensen. A common drawback of these studies is that a given author rarely dissected and analysed several species of a single taxon using one consistent method. Only in very few cases have histological sections, dissected specimens and so on, been preserved. The conclusions drawn from the original data can therefore not always be scrutinized. Furthermore, the results were sometimes biased by prevailing morphological hypotheses. For example, recent histological studies [60] as well as our findings (Figure 3) indicate that the siphon, a canal bypassing the sea urchin stomach that was believed not to be found in the Diadematoida (see, for example, [61]), is indeed also present in this group of animals. Distinct features of internal sea urchin anatomy may extend to the sizes and shapes of digestive tract, gonads, ampullae and muscles. A thorough revision of internal structures in the light of recent technological advances will therefore almost certainly change our view of facts that, for years, were taken for granted. Scanning of sea urchins has so far been performed by us on more than 50 species (Ziegler et al., publication in preparation).
The feasibility of comparative morphological studies using MRI is further exemplified by our description of the highly variable shape and size of the gastric caecum found in the Irregularia ( Figure 5). Our findings were correlated with the little information available on this structure; the combined data led us to assume that a reduced gastric caecum consisting of several smaller blindly ending sacs might constitute a soft tissue character common to all cassiduloid sea urchins. We therefore predict the presence of a similarly shaped structure in all species of the Echinolampadidae and Cassidulidae. Furthermore, it would be of interest to extend our analyses to members of the Apatopygidae and Neolampadidae, the other two extant families currently assigned to the Cassiduloida.
Although for some organs high intraspecific variability was detected, in particular in the case of organs with a development-dependent size and shape such as the gonads ( Figure 1E and F), shape and localization of other organs (for example the digestive tract) did not differ significantly among specimens from a given echinoid species, allowing comparison on various taxic levels (Figures 3 and 4). However, intraspecific variability of internal organs can also be observed in other taxa, for example gastropods (sinistral freak) or humans (situs inversus viscerum). Therefore, like larval and hard-part morphology, which are continuously updated [62,63], morphological character sets derived from internal organs of sea urchins will serve as a useful systematic reference for future studies.

Non-invasive imaging techniques other than MRI
A number of imaging techniques allow for non-invasive morphological studies on whole specimens. However, only a few of them are able to achieve resolutions that permit detailed morphological analyses in small animals. Confocal laser scanning microscopy (CLSM), in combination with the autofluorescence of the crustacean or insect cuticula, can be used for 3D reconstructions of minuscule body parts or whole specimens [64][65][66]. The potential of this promising technique must not be underestimated, since the achievable resolutions permit taxonomic studies (based, for example, on chaetotaxy) as well as large-scale systematic morphological analyses. However, this procedure reveals internal structures only in rare cases and is limited to tiny objects.
In contrast to CLSM, µCT has already become a standard tool for detailed non-invasive studies of whole specimens, especially in studies on fossils [67]. It is based on the analysis of a sample by means of X-rays and can primarily be used to reveal hard tissue structures such as bone or calcite. Moreover, recent studies using more advanced techniques such as X-ray synchrotron microtomography and a new setup called very-high-resolution X-ray computed tomography have been employed to display soft tissues, in vitro as well as in vivo [6,7,[68][69][70] (see also the review by Attwood [4]). Using µCT, isotropic resolutions of less than (5 µm) 3 have been achieved on whole specimens, although, technically, resolutions in the nanometre scale are within reach. However, whether these new µCT methods can compete with the ability of MRI to display soft tissue at high resolutions still needs to be evaluated.
Our own preliminary studies to generate images of internal organs performed on several sea urchin museum specimens (Echinoneus cyclostomus Leske, 1778, Psammechinus miliaris (Müller, 1771), Mespilia globulus (Linnaeus, 1758), and Moira atropos (Lamarck, 1816)) using a desktop high-resolution µCT scanner (Skyscan 1172, Skyscan, Kontich, Belgium) at the Max-Planck-Institut für evolutionäre Anthropologie, Leipzig, Germany were not successful, although these µCT datasets can be used for the reconstruction of hard tissue structures. 3D models based on this data will be made available on the DigiMorph website These results demonstrate that complementary imaging techniques should ideally be used for comparative morphological analyses, an approach commonly referred to as multimodality.

Conclusions
Our results extend the use of MRI in invertebrate morphology to a systematic approach using museum specimens of a major marine invertebrate taxon, the sea urchins (Echinoidea). MRI can be employed for the rapid, non-destructive and unbiased acquisition of digital morphological data, and the creation of interactive 3D models accessible both within a publication and via the Web. The revelation of the internal anatomy of echinoid museum specimens at a resolution comparable with manual dissection extends our knowledge of soft tissue characters, permitting phylogenetic inferences.

Specimen preparation
All specimens analyzed belong to the taxon Echinoidea (sea urchins), one of the five subtaxa of the Echinodermata (marine spiny-skinned animals). The museum specimens were mostly fixed in formalin and all were conserved in alcohol, although in some cases the exact mode of fixation could not be determined. For MR scanning, the museum specimens had to be lowered down to distilled water in a gradual alcohol series. Specimens that were obtained in living state were fixed in a 7% formaldehyde solution. For MR scanning, the freshly fixed specimens were kept in this solution. Results derived from specimens scanned in alcohol and formalin did not appear to differ significantly.
The specimens where placed either inside a custom-built Plexiglas chamber (Berlin), inside a 50 ml Falcon tube (Berlin), or inside NMR tubes with diameters between 5 and 20 mm (Würzburg), depending on the size of the specimen. On some specimens spines had to be dressed for tight fit inside the tubes. Magnevist (Bayer-Schering, Berlin, Germany), a gadolinium-based non-selective contrast agent, was added at a final concentration of 2 mM. This concentration had been employed successfully in preliminary studies (results not shown).
The addition of the contrast agent to the museum specimens was discussed beforehand with the curators who judged the risk to the specimens' integrity as negligible. Prior to scanning, samples were degassed at 50 mbar. The freshly fixed specimens used in this study will be deposited as voucher material at the ZMB. Museum specimens will be stored in separate jars for potential later re-scanning. Table 2 lists detailed information on every museum and freshly fixed specimen used in this study. Scanning protocols employed a 3D gradient echo imaging sequence (FLASH) with parameters depending on the specimen size (see Additional file 2 for details). Almost all scans resulted in datasets with isotropic spatial resolution.

Specimen scanning and photography
The photos of museum and freshly fixed specimens were made using a digital camera (Casio Exilim) with 7.2 Megapixels. The acquired images were processed using Adobe Photoshop PDF. The 3D model was embedded into this publication using the Adobe 3D Toolkit according to the procedures described in [71].

51.
The         The gastric caecum is a translucent body free of sediment and probably constitutes one of the main sites of digestion [38,39]. The upper diagrams show an aboral view, the lower diagrams a lateral view with the anterior side (ambulacrum III) oriented towards the right-hand side.
Arrows  1 cm. The colour legend specifies organ designation. The interactive 3D mode can be accessed by clicking onto Figure 6 in the 3D PDF version of this article: Additional file 7 (Adobe Reader Version 7.1 or higher required).

Additional files
Additional file 1 Table1.pdf PDF file