The Potent Respiratory System of Osedax mucofloris (Siboglinidae, Annelida) - A Prerequisite for the Origin of Bone-Eating Osedax?

Members of the conspicuous bone-eating genus, Osedax, are widely distributed on whale falls in the Pacific and Atlantic Oceans. These gutless annelids contain endosymbiotic heterotrophic bacteria in a branching root system embedded in the bones of vertebrates, whereas a trunk and anterior palps extend into the surrounding water. The unique life style within a bone environment is challenged by the high bacterial activity on, and within, the bone matrix possibly causing O2 depletion, and build-up of potentially toxic sulphide. We measured the O2 distribution around embedded Osedax and showed that the bone microenvironment is anoxic. Morphological studies showed that ventilation mechanisms in Osedax are restricted to the anterior palps, which are optimized for high O2 uptake by possessing a large surface area, large surface to volume ratio, and short diffusion distances. The blood vascular system comprises large vessels in the trunk, which facilitate an ample supply of oxygenated blood from the anterior crown to a highly vascularised root structure. Respirometry studies of O. mucofloris showed a high O2 consumption that exceeded the average O2 consumption of a broad line of resting annelids without endosymbionts. We regard this combination of features of the respiratory system of O. mucofloris as an adaptation to their unique nutrition strategy with roots embedded in anoxic bones and elevated O2 demand due to aerobic heterotrophic endosymbionts.


Introduction
The unique bone-eating organism, Osedax (Siboglinidae, Annelida) was first described in 2004 from a whale fall located at 2891 m depth in Monterey Bay, Pacific Ocean [1]. Since its first discovery it has been found on multiple whale falls in the Pacific and Atlantic Oceans, artificially deployed cow bones [2] as well as on other vertebrate bones such as those of teleost [3]. There are five formally described species, with at least a further 12 species known from genetic evidence [4,5]. In addition, convincing fossil traces of Osedax have been found in Oligocene and Pliocene mammal bones [6,7].
Evidence suggests that Osedax females utilize the complex organic compounds of the bone through endosymbiotic aerobic heterotrophic bacteria (Oceanospirillales) located in bacteriocytes in a root system that is embedded in the bone matrix [8,9]. The O 2 microenvironment around the embedded root system of Osedax has not been studied, yet this knowledge is crucial for understanding the function of Osedax in its natural habitat. The O 2 supply within the bone matrix is presumably strongly diffusion limited, and with whale bone lipid content reaching 45% in Sei whales (closely related to Minke whales) [10] the bone interior may become O 2 depleted due to high heterotrophic microbial activity, including sulphur-reducing bacterial processes that generate sulphide, further reducing O 2 below the bone surface. External O 2 supply to the roots of Osedax via the bone matrix is thus highly unlikely. Yet, symbiosis with detoxifying sulphide-oxidizing bacteria has so far not been proven, though Osedax has been observed along with sulphophilic species and mats of white Beggiatoa-resembling bacteria (Figure 1) or bones exhibiting ferrous sulphide precipitation [11][12][13][14]. The highly folded epidermis of the root structure of Osedax 'green palp' has recently been shown to lack a cuticle and possess an extensive microvillous border [15], potentially facilitating the uptake of organic substrates, but also facilitating interaction with sulphide from the root surroundings. In the methane seep-dwelling vestimentiferan Lamellibrachia the sediment-embedded posterior body region (also called root) is actually the main source of hydrogen sulphide uptake, and is used for maintaining the chemoautotrophic endosymbionts [16,17]. Respiratory features in other siboglinids involve ciliary or muscular ventilation of the chitinous tube and expanded branchial structures with extensive blood vascular systems and large respiratory surface areas [16,[18][19][20]. Whereas the tube of Osedax is gelatinous and only covers the trunk, the vascularized and pinnulated palps of Osedax are believed to have a similar function as the anterior branchial plumes of Vestimentifera [21]. However, neither these structures nor the blood vascular system have been examined in detail, nor has the O 2 consumption been measured previosly in any Osedax species.
Osedax and Vestimentifera are tubicolous species and apart from the ability to retract into their tubes when disturbed, they show very little motile activity. However, Riftia pachyptila has been shown to metabolize much faster than other resting annelids (e.g. double the O 2 demand of Arenicola marina Lamarck, 1801) [22]. Moreover, Childress et al. [22] apparently measured the O 2 uptake without the presence of sulphide, thereby underestimating the O 2 consumption of chemosynthetic endosymbionts. Later, Girguis & Childress [23] showed that O 2 uptake was reduced significantly (from 14.3561.23 mmol O 2 g 21 wet wt h 21 to 2.8860.89 mmol O 2 g 21 wet wt h 21 ) when the activity of endosymbionts was restricted by blocking sulphide supply. This clearly demonstrated that the metabolism of the endosymbionts of R. pachyptila contribute significantly to its O 2 consumption. Experiments by Freytag et al. [17] further confirms this as an increase in O 2 consumption of ,42% was seen when Lamellibrachia cf. luymesi was exposed to hydrogen sulphide at the root area. The increase in O 2 consumption is a result of the activation of the metabolism of chemoautotrophic endosymbionts. Since Osedax also carries aerobic endosymbionts and furthermore is hypothesised to be partly exposed to hypoxia or anoxia, knowledge of the O 2 uptake of Osedax as compared to R. pachyptila and other annelids is relevant for enhancing our understanding of the respiratory systems and physiology of Osedax.
The shallow water species Osedax mucofloris Glover, Kä llström, Smith, Dahlgren, 2005, has been found at 30 m and 125 m water depth off the coast of Tjä rnö, Sweden [12,13], and at 120 m depth in Bjørnafjord, Norway [24]. Currently these are the only records of Osedax from the Atlantic, though other species are found in shallow waters in both the East and West Pacific [4,25]. The 125 m deep locality near Tjä rnö is characterized by stable oceanic salinity (34-35%), temperature (5-7uC) and dissolved O 2 concentration ranging from 4.6 to 6.3 ml O 2 l 21 [13].
In the present study we investigated whether Osedax mucofloris is exposed to an inhospitable hypoxic or anoxic microenvironment by measuring the O 2 distribution surrounding the embedded root system. We assessed possible morphological and/or physiological adaptations to the environmental conditions through detailed morphological studies of the respiratory surfaces and the blood vascular system as well as respiratory measurements of O 2 consumption. The results are discussed in relation to the unique environment, endosymbionts, and embedded root structure of Osedax as well as compared with studies of related annelids.

Ventilating mechanisms
Ciliary bands and sensory structures (anti acetylated atubulin staining and histology). In the description below, we use a dorsal-ventral definition opposite to the one used by Rouse et al. [1,21], based on a new interpretation (Rouse & Worsaae, unpublished).
We found two previously unreported broad longitudinal ciliary bands dorso-laterally on each side of the oviduct on the anterior part of the trunk (Figure 2A, B, E). The ciliary bundles of the bands consist of multi ciliated cells, appearing pillow-like with numerous, conspicuously short cilia (,10 mm long) projecting outwards from the centre of the elliptical bundles ( Figure 2C). The bundles are organized in an anterior-dorsal diagonal pattern within the longitudinal bands ( Figure 2B). The ciliary bands narrow toward the posterior part of the trunk and are replaced by single tufts of cilia scattered basally across the trunk surface ( Figure 2A). The short cilia appeared immotile, and with several longitudinal nerves running beneath, (Worsaae & Rouse, unpublished) their function may be sensory rather than ventilatory.
Two dense ciliary bands are found on each palp along their epidermal longitudinal lobes on each lateral side, as previously reported [12] (Figures 3A, B, E, 4C, 5A, F, G). The bands consist of multi ciliated cells with 70-90 mm long cilia ( Figure 3E). The ciliary bands extend from near the basal part of the palps to the distal tip, beating in metachronal waves.
Two nerves, originating at the anterior part of the brain, innervate each palp. One nerve runs abfrontally, between the two lateral ciliary bands and the other nerve runs laterally underneath one of the ciliary bands ( Figure 5A). Further details on the female Osedax nervous system will be described elsewhere (Worsaae & Rouse, unpublished).
Pinnules (up to 100 mm wide) project perpendicularly from the frontal palp surface between the lateral ciliary bands, with a density of approximately eight pinnules across the palp per 50 mm palp length (Figures 3A, 5F, G). At the proximal end of the palps, pinnules are less developed than at the distal end, seemingly growing in length (up to 170 mm) synchronously with the growth of the palp ( Figure 3A). A sensory cell extends through the centre of the pinnule with a distal perikaryon and a few external, presumably sensory, cilia ( Figure 4C, D). Its axon seems to connect with one of the two major longitudinal palp nerves, possibly the  Single z-stack image of the trunk musculature, longitudinal muscles beneath circular and diagonal muscles. Abbreviations: ciliary tufts (ct), circular muscles (cm), diagonal muscles (dm), elliptical ciliary bundles (ecb), lateral ciliary band (lcb), longitudinal muscles (lm), muscular gap (mg), palp (p), root structure (r), torn ovisac (to), trunk (t), dorsal blood vessel (dbv). doi:10.1371/journal.pone.0035975.g002 one running more laterally beneath the ciliary band. This is supported by similar findings in Osedax ''yellow-collar'' ( Figure 4E).

Musculature
(F-actin, phalloidin staining). The longitudinal muscles run along the entire length of the trunk (Figure 2A), originating posteriorly at the trunk basis, and inserting anteriorly at the base of the four palps ( Figure 3D). In one individual, it was possible to detect clustering of longitudinal muscles into 14-16 bundles of .20 muscle strands in the posterior part of the trunk, anterior to the root structure ( Figure 4A). Along, and around the entire trunk, the muscles are distributed in a dense cylindrical formation with an average of six strands per 50 mm ( Figures 2D, E, 3D). The musculature is slightly separated internally to the oviduct (possibly by the nerve cords), as well as randomly along the trunk, creating minor gaps most likely for mucus gland exits or nerves (Figures 2A, 3D). Anteriorly, a gap in the musculature is found ventrally at the position of the brain, as well as at each of the four insertion points of the palps. At each of these four points, the longitudinal musculature divides into two bundles (of each 20-30 strands), encircling the insertion point of the palp ( Figure 3D). The longitudinal palp muscles originate at the base and run along the entire length of the palps to their tips, separated by a smaller frontal and a larger abfrontal gap ( Figures 3C, D, 5A). No longitudinal musculature was detected in the pinnules.
Around the ovisac and anterior root structure the longitudinal musculature divides, and together with the circular muscles, creates a mesh-like structure ( Figure 4A). The musculature extends posteriorly along the roots in a cylindrical formation, supporting the tissue penetrating the bone.
Thin circular muscles are found peripheral to the longitudinal muscles along the entire trunk, with 24 strands per 50 mm ( Figure 2D). The circular muscles are most dominant in the posterior part of the trunk and continue into the root structure ( Figure 4A). Notably, much thicker diagonal muscle strands were found beneath the circular musculature, but peripheral to the longitudinal muscles ( Figure 2F). Attached at the mid-dorsal line, the strands run diagonally around the trunk in an anterior direction and attach at the mid-ventral line. The diagonal muscles are distributed along the entire length of the trunk, lying further The circular musculature of the palps encircles the longitudinal musculature as a cylinder with ,18 muscle strands per 50 mm and is evenly distributed along the entire palp ( Figure 3C). Fine circular muscles enclose each vessel of the pinnular loop ( Figure 4C, D); they were likewise visible in the semi-thin sections on the outside of the pinnular loop and along the midline of each pinnule ( Figure 5F, arrow tips), with a spacing corresponding to those shown with phalloidin staining ( Figure 4C, D).

Blood vascular system
The trunk of Osedax mucofloris encloses two major longitudinal blood vessels ( Figures 2E, 6A, B). The dorsal vessel is highly muscularized with circular musculature throughout its length ( Figure 2D, E). No lateral connecting vessels between the major longitudinal vessels or epidermal capillaries were found in the trunk. From the trunk, the two blood vessels continue posteriorly into the root structure and increase in diameter ( Figure 6A, B). When studied under the light microscope, the muscularized dorsal vessel is visible as a defined tube along the trunk, and continues into the anterior root structure, curling up in the centre anterior to the ovisac. The curling configuration is most likely caused by contraction of the basal part of the trunk. The exact further path of the vessel was difficult to determine. The Confocal Laser Scanning Microscope (CLSM) studies also showed the continuation of the dorsal blood vessel around the ovisac, revealing cylindrical musculature at the ovisac, corresponding in diameter to the musculature of the dorsal blood vessel of the trunk ( Figure 4B).
The ventral blood vessel also continues into the root structure, but as a narrower and less defined vessel, the path of which was even more difficult to determine, than that of the dorsal vessel. Blood vessels of different sizes, as well as multiple obvious capillaries within the root tissue were observed ( Figure 6B-E). Larger vessels extend out from the area of the ovisac and divide into thinner vessels (diameter: 7-30 mm). Capillaries were detected in the periphery of the root tissue ( Figure 6E), with distances between the detected capillaries ranging from 65-220 mm.
Anteriorly, each palp encloses a pair of blood vessels created by invaginations of the inner lamina of the basement membrane of the epidermis (Figures 5F, G, 7C). Imaging of live specimens confirmed the presence of two blood vessels, running along the entire length of the palps.
The pinnules are largely filled by a blood cavity lined with a membrane, which fuses in between the two blood cavities along most of the pinnule length, thereby creating the pinnular loop ( Figures 5B-G, 7C). The palps of the histological sections were ,300 mm in diameter at the base and the palp epidermis was ,50-75 mm thick. The palp diameter and the thickness of the Pinnules vary in diameter along their length and along the palp, with a median diameter of ,406100 mm for the pinnules and ,20-40 mm for each blood vessel. The diffusion distance across the pinnule epithelium was measured on the semi-thin sections to be 1-2 mm, and for the epidermis of the distal part of the palps, the diffusion distance was measured to be ,30 mm.

Oxygen consumption
When corrected for background respiration, the measured O 2 consumption (MO 2 ) of O. mucofloris ranged almost across a factor of ten from 220693 mg O 2 g 21 h 21 to 205361950 mg O 2 g 21 h 21 (Table 1)

Oxygen levels around Osedax
Micro sensor measurements of O 2 concentrations in proximity to the bone interface were conducted, through agar-filled holes, on Osedax-colonized bone fragments in cuvettes. The obtained profiles of O 2 concentration from the aerated seawater, and inwards showed steep O 2 gradients towards both the bone and tissue surface ( Figure 8A, B). Anoxic conditions or very low O 2 levels were found at the bone surface, within the bone, and in proximity to the embedded root tissue of O. mucofloris ( Table 2). The average Micro sensor measurements of O 2 distribution in one mucus tube showed a ,50% decrease in the O 2 concentration in the centre of the mucus tube wall as compared to outside the tube (Table 3). Direct measurements within the tube were not possible due to disturbance by the worm. Measurements of the O 2 microenvironment surrounding the palps showed a strong decrease in O 2 concentrations, when a palp approached the microelectrode measuring tip. At the base and middle of the palp, the O 2 levels were almost zero showing the O 2 uptake to be high in these areas (Table 3). At the distal end of the palp the O 2 concentration was only reduced to approximately 50% atmospheric saturation, possibly due to decaying palp tips.

Ventilation and branchial structures
The presence of highly vasculated palps and pinnules, the former densely ciliated showed that the anterior crown is the main site for O 2 uptake ( Figure 7A-C). Uptake of O 2 over the trunk surface is possible, but a thicker epidermis, short and seemingly immotile ciliary bands and no obvious respiratory structures suggest that the trunk is a minor site of O 2 uptake. Oxygen does, to some extent, diffuse from the surrounding water into the mucus tube, thereby supplying the dwarf males.
The well-developed longitudinal musculature of the trunk serves to retract the trunk and palps into the tube and bone, presumably as protection from predators. As no regular retraction patterns of trunk musculature were detected, retraction into the tube is not considered a significant mode of ventilation for required O 2 . Furthermore, the tube only surrounds the trunk, tightly fitting to the base of the trunk and the surface of the bone, preventing any water exchange to the ovisac and roots from the bone surface. The longitudinal musculature is moreover able to stretch the extensive branchial structures (palps) into the aerated water in order to increase the O 2 uptake.
The circular trunk musculature in O. mucofloris is weakly developed, as found in several other annelids [26,27]. The diagonal musculature revealed in the present study, may compensate for the weak circular musculature by having a similar supportive function. Diagonal musculature has most likely been mistaken for circular musculature in several annelids [26,27], including previous studies of Osedax [15,21].
Our study demonstrates the respiratory challenges faced by Osedax, as no oxygen was measured in the bone matrix as well as in the immediate vicinity of root structures and ovisac. Furthermore, we found no ventilating structures such as ciliary bands on the root epidermis, nor dense capillaries beneath the epidermis. Therefore, the epidermis of the embedded root structure is unlikely to facilitate ventilation or local O 2 supply to the Osedax root system. The O 2 necessary to maintain the metabolism and heterotrophic endosymbionts of the roots may instead primarily be transported via the blood vascular system from uptake at the aerated palps ( Figure 7A). However, the microvillar root epidermis in Osedax does constitute a large surface area to volume ratio, which besides uptake of organic compounds may also facilitate uptake of e.g., hydrogen sulphide in the bone matrix. This is the case for the 'root' surface of the vestimentiferan Lamellibrachia, where hydrogen sulphide is absorbed and then transported via the circulatory system to the chemoautotrophic endosymbionts in the trophosome [16,17,28].
Previous studies of other close relatives of Osedax such as Vestimentifera and Sabellidae have shown anterior branchial structures to be of vital importance when tube-bound tissue is not Osedax mucofloris extend its palps and pinnules, with large respiratory surfaces, into the overlying O 2 -rich water in order to uptake O 2 . O 2 is then distributed to the buried root system through the extensive blood vascular system also supplying the heterotrophic endosymbionts. The O 2 distribution to the root system is crucial as local uptake is not possible in the anoxic bone environment. The anoxic environment is partly produced by intense bacterial processes (green arrows) utilizing O 2 at the bone surface. Hydrogen sulphide is produced by anoxic bacterial processes within the bone matrix during decomposition of organic content using sulphate. B: Schematic illustration of assumed blood flow in palp and pinnules, longitudinal section. Blue vessels carrying venous blood through afferent vessels, red vessels carrying arterial blood through efferent vessels. C: Schematic illustration of assumed blood flow in palp and pinnules, transverse section. Likewise blue vessels carries venous blood through afferent vessels, red vessels carries arterial blood through efferent vessels. Note that the palp blood vessels are created by an invagination of the basement membrane. Green indicates musculature. doi:10.1371/journal.pone.0035975.g007 ventilated [29][30][31]. Likewise, the anterior branchial structures of O. mucofloris seem morphologically adapted to facilitate efficient O 2 uptake form the surrounding water. As with the branchial plume of Vestimentifera [31,32], the anterior palps and pinnules of O. mucofloris have a large surface area to volume ratio and short diffusion distances. Furthermore, O 2 uptake is optimized by the two ventilating ciliary bands on each palp, a feature also seen in other Siboglinidae and in Sabellida in general [20,31,33]. A rough estimate was made of the surface area of the anterior crown, calculated from palp length given by Glover et al. [12], and the dimensions of palps and pinnules of O. mucofloris found in the present study. This results in a weight specific branchial surface area (SBSA) of ,22 cm 2 g 21 fixed mass, which is similar to the SBSA of Riftia pachyptila and higher than the SBSA of fish, crabs and other annelids [31]. This is especially obvious when compared to the SBSA of e.g., Arenicola marina (4.00 cm 2 g 21 ) [34], although this species can also take up O 2 across its general epidermis.
Interestingly, the present study showed that each pinnule is equipped with a distal sensory cell and external sensory cilia. These structures may sense disturbance in the water to avoid predators, as Osedax need not sense food items or reproductive indicators in the water [1,8,35]. Additionally, sensing of currents may also be beneficial in order to orientate the palps e.g., for optimal uptake of dissolved O 2 . Riftia pachyptila does not have sensory structures on the pinnules, but does have long, separate sensory filaments that lack ciliary bands and pinnules. These structures, with unknown function, are placed between pinnulated filaments [33]. The quantity and complexity of capillaries in the root structure of O. mucofloris reflects the high O 2 demand of Osedax, presumably for the metabolism of its heterotrophic endosymbionts and production and development of eggs. Similar capillaries are visible on the exposed ovisac of O. frankpressi (figure 2F in [1]) and O. roseus (figure 4E in [21]), while the present study shows the presence of capillaries supplying the more distally placed root tissue and bacteriocytes. However, the extent of capillaries does not match the extensive capillary network of the trophosome in other Siboglinidae [20,33]. This is in accordance with the original description [1] mentioning the lack of a discrete trophosome, the Osedax trophosome instead being diffuse and beneath the epidermis of the embedded tissue.
The present study supports the basic histological findings with regard to the circulatory system of the palps in O. roseus [21]. The increased level of detail in the present study suggests similarities with some previous siboglinid studies [31,32,36]. The pinnular loop is most likely lined by an extension of the outer lamella of the basement membrane, which is in accordance with the findings of Nørrevang [36]. No indication of a capillary plexus between the two elements of the loop was observed, as otherwise seen in Riftia pachyptila and Ridgeia piscesae [31][32][33]. In accordance with Nørrevang's [36] observations of blood flow in Siboglinum, the blood of Osedax flows through an afferent vessel into the palp, through the pinnular loops and back through an efferent palp vessel (Figures 7B-C). This circulatory system carries oxygenated blood to the trunk and root system. In the present study of O. mucofloris and previous studies of Osedax [1,21] the main trunk blood vessel has been shown to be muscularized, and is now regarded as dorsal (Rouse & Worsaae, unpublished). This is in accordance with other annelids [37] and siboglinids [20,33], where the muscularized dorsal vessel, possibly assisted by a heart, create blood flow in an anterior direction in the dorsal trunk vessel and posterior in the ventral trunk vessel. Rouse et al. [1] reported an anterior lying dorsal heart in the original description of the genus, not noted in any descriptions since and would now appear to be an error (Rouse, pers. obs.).
The main blood flow within the palp vessels may be generated by the circular body wall musculature of the palp. However, the thin circular musculature of the pinnules surrounding each branch of the looped blood vessel ( Figure 4C, D) most likely assists local blood flow, as suggested for the tentacular vessels of Riftia pachyptila [22]. Musculature in anterior appendages has been found in several Vestimentifera, in the form of sphincter muscles located in the branchial lamella and filament vessels [33].
As with the trunk of O. mucofloris, the anterior vestimentum of Vestimentifera lacks branching vessels between the ventral and the muscularized dorsal vessel [33]. The resemblance between the blood vascular systems adds new evidence to the hypothesis that the trunk of Osedax and the vestimentum of Vestimentifera are homologous regions [21].

Oxygen consumption
The present study shows that Osedax mucofloris has a higher weight specific O 2 consumption (MO 2 ) than other resting annelids ( Figure 9) [18]. This may reflect an elevated demand of the embedded tissue due to presence of heterotrophic aerobic endosymbionts, which have a higher metabolism than regular tissue. The measured MO 2 actually corresponds to that found for Riftia pachyptila [22,23] (Figure 9), possessing a vast amount of chemoautotrophic bacteria in their trophosome. Furthermore, a high MO 2 may also reflect oxidative sulphide detoxification. The high MO 2 corresponds well with the large SBSA and elaborate branchial structures, which are both usually associated with animals exhibiting a high O 2 demand. This correlation is also found in Arenicola marina with smaller SBSA and lower MO 2 (red square, Figure 9).
Our data show that O. mucofloris inhabiting bone in cuvettes (purple and yellow dots, Figure 9) have a lower MO 2 than O. mucofloris inhabiting sectioned cow bone (green, blue, red dots, Figure 9). Along with high standard deviations, this highlights the difficulties of measuring O 2 consumption on these embedded worms. The variations are most likely caused by the difference in blind respiration measurements and the large biological activity present on decaying bone, which was difficult to quantify. Future studies should thus be carried out to more precisely determine the

Osedax adaptations to the bone environment
The bone matrix exhibited strong O 2 depletion with anoxic or low O 2 levels at its surface and with areal O 2 consumption rates comparable to in situ diffusive O 2 uptakes measured in sediments proximate to whale falls [38]. The respiratory and circulatory system of Osedax is well suited to these environmental challenges. The elaborate branchial structures of the palp and pinnules of Osedax mucofloris facilitate a high O 2 uptake from the surrounding seawater that, in combination with the blood vascular system, apparently enables sufficient oxygenation of the embedded tissue and endosymbionts. A similar, well-developed respiratory system is also found in other members of Siboglinidae (for review see [39]), also exhibiting high surface area to volume ratio in the branchial structures in particular. Based on these morphological and functional similarities among several siboglinids it seems possible that a well-developed anterior branchial structure may be a plesiomorphic condition among siboglinids and possibly a prerequisite for housing aerobic symbionts while dwelling in anoxic habitats. This respiratory system may therefore very well be a preadaptation present in the common ancestors of Osedax and other siboglinids, responsible for the successful colonization and evolution of Osedax in the unique bone environment.
An extended root surface (and high area to volume ratio) is likewise found in both Lamellibrachia (e.g., [16]) and Osedax (yet branched). However, the main function has so far been interpreted as very different. Whereas the Osedax root surface is suggested to mainly facilitate uptake of organic compounds (with less focus on the possible congruent uptake of sulphide) [9], the root of Lamellibrachia is found to be the main respiratory surface of hydrogen sulphide necessary for the chemosynthesis of the symbionts [16,17,28].
The measured O 2 depleted bone environment as well as ferrous sulphide precipitations ( Figure 8E) supports the suggested exposure of Osedax to sulphide. Furthermore the large root surface in the related Lamellibrachia is highly efficient in hydrogen sulphide uptake [17]. Since the main trophic source of Osedax compared to other siboglinids (including Lamellibrachia) seems to be heterotrophic bacteria rather than sulphide-detoxifying chemoautotrophic bacteria [8], Osedax may instead possess physiological adaptations to detoxify sulphide.
Though beyond the scope of the present paper, microanalytical approaches such as microsensors [40,41] and functional imaging techniques [42] could yield a more complete mapping of the chemical microenvironment of Osedax, including the exact levels of sulphide exposure and spatio-temporal dynamics of O 2 in the root system. It would also be interesting to know whether Osedax have sulphide-binding properties of their haemoglobin as found in Vestimentifera [43,44], as they might use them to transport toxic sulphides from the area surrounding the root structure to e.g., the palps, somehow releasing the toxic compounds to oxygenated seawater or detoxifying them.

Sampling and aquarium setup
In January 2009, 3 replicate experimental sampling devices, using cow and whale bone, for recruitment of female O. mucofloris ( Figure 10) were placed at 125 m depth off the coast of Tjä rnö, Sweden (58u52.976N; 11u05.715E) in close vicinity to a minke whale carcass sunk in October 2003 [13]. One device for morphological and reproductive studies was successfully retrieved in May 2009. A second device was retrieved in November 2009 for in vivo studies (respirometry and microsensor analysis) and additional morphological studies. Additionally, a piece of cetacean Triangles: Previous measured O 2 consumption of R. pachyptila. No sulphide present in water when measuring: Red, blue [22] and green [23]; sulphide present in water during measurement: Purple [23]. Red Square: O 2 consumption of resting Arenicola marina [52]. doi:10.1371/journal.pone.0035975.g009 bone deposited at 123 m in the same area in May 2008 was retrieved in January 2009 and used for preliminary investigations and for designing experimental setups. In January 2009, the bottom water had a salinity of 34.6%, a temperature of 8.4uC, and an O 2 content of 78.7% atmospheric saturation.
Placement and retrieval were carried out with a remotely operated vehicle (ROV, Sperre Subfighter, Norway) operated from the R/V Lophelia (Sven Lovén Centre for Marine Science, University of Gothenburg, Sweden). Upon retrieval, the sampling device was placed in a cooling box with ambient aerated water and transported to the Marine Biological Section, University of Copenhagen within 8 hours. In the laboratory bones were kept in aerated seawater under in situ conditions, in aquariums (Vol 60 litre) with aeration, protein skimmers and a recirculating pumping system. The Osedax ''yellow-collar'' shown as comparison in Figure 4 was sampled in November 2009 in Monterey Bay, California at 385 m from a Grey whale skeleton and stained in accordance to the protocol described below. All necessary permits were obtained for the described whale fall experiment; received from Karin Pettersson at the County Administrative Board Vä stra Götalands Lä n, Oct. 2003. The field studies did not involve endangered or protected species.

Fixation
Specimens were carefully dissected from the bone and fixed for immunohistochemistry, benzidine staining and histology. Prior to dissection, Osedax mucofloris were anesthetized for 5-10 minutes in a 1:1 solution of seawater and MgCl 2 (isotonic to seawater). Anaesthetized animals were gently dissected with scalpels and fixed at 4uC over night in 4% paraformaldehyde in 0.15 M phosphate-buffered saline (PBS) with 5% sucrose, pH 7.4. Subsequently, the animals were rinsed 4-6 times for 30 min in PBS with 5% sucrose and were then stored at 4uC in PBS with 0.05% sodium azide (NaN 3 ).
Specimens were studied using a Leica TCS SP5 confocal laser scanning microscope (CLSM) (University of Copenhagen, Faculty of Health Science, courtesy of M. Givskov and T. Bjarnsholt). Leica LASAF computer software or ImarisH x64 6.0.0 (Bitplane AG, Zurich, Switzerland) was used to produce projections of zstacks of CLSM images, while further analyses of z-stack series were performed with ImarisH x64 6.0.0. Computed 2D images of muscles, nerve and cilia with relation to respiration and palp morphology were further optimized with Adobe Photoshop CS3 and Adobe Illustrator (Adobe System Incorporated) for presentation.

Histology
One specimen was embedded in epon and used for histological analysis. Semi-thick 1.2 mm sections were cut on a microtome (EM UC6, Leica, Wetzlar Germany) with a diamond knife (Diatome; Biel, Switzerland). A small amount of Pattex contact adhesive (Pattex Compact; Henkel KGaA, Düsseldorf, Germany) was diluted with a few drops of xylene in an Eppendorf tube and applied to the side of the epon block to make serial sectioning of ribbons possible following the protocol of Henry [46] and Ruthensteiner [47]. Bands of ,20 sections were stained with toluidine blue and mounted in EntellanH (Electron Microscopy Sciences, Pennsylvania, USA). Sections were studied and photographed using light microscopy (BX50 microscope; DP71 camera; Cell F software; Olympus, Japan).

Benzidine staining of haemoglobin
Using a modified version of the benzidine staining method by Knox [48], haemoglobin was stained in four fixed specimens. A 100% saturated benzidine solution was prepared by adding benzidine to distilled water. The solution was stirred for two hours. Fixed specimens were rinsed in running tap water in the same time period. Specimens were subsequently incubated in the filtered benzidine solution for 1 hour, also under stirring. Next, 3% hydrogen peroxide was added drop by drop until blood vessels turned dark blue. Specimens were either mounted in glycerol directly or dehydrated in a series of alcohol acidified with drops of 0.1% acetic acid, where after tissues were cleared in xylene and mounted in D.P.X between two cover slips. Specimens were analyzed and photographed under a light microscope (BX50 microscope; Dp71 camera; Cell F software; Olympus, Japan).

Oxygen consumption measurements
O 2 consumption was measured in seawater kept at ,100% atmospheric saturation using intermittent respirometry in accordance with Vismann and Hagerman [49]. The experimental setup was placed in a constant temperature room at 6uC. Each experiment encompassed 3-8 measuring sequences consisting of a flushing period of 10-30 min and a measuring period of 30-45 min. The set-up had a chamber flushing rate of 30 ml min 21 and a shunt water flow of 8 ml min 21 where pO 2 = the O 2 partial pressure at full saturation (kPa), ww = wet weight of fixed specimen rinsed for mucus tube (g) (tissue fixed in 4% parafomaldehyde and stored in PBS is assumed to have similar weight as unfixed tissue), a = slope of regression line for O 2 decrease (% h 21 ), b = O 2 solubility (mg O 2 l 21 kPa 21 ) and v = volume of the respiration chamber (l). O 2 consumption was measured on O. mucofloris inhabiting three sectioned cow bone pieces (B1, B2 and B3) and bones in two cuvettes (C1 and C2). Measurements were either initiated directly after the annelids protracted subsequent to the disturbance of being moved (C1, C2, B1) or after one night of acclimatization (B2, B3). All measurements were corrected for background respiration. For B2 and B3 the background respiration was measured using the same water and bones (after dissection and 48 hours of acclimation to restore biological activity). For C1, C2 and B1 the background respiration was the mean value of measurements using new water, bones in two cuvettes and three sectioned bone pieces without O. mucofloris. Different respiration chambers were used for sliced bone pieces (volume: 186 ml) and cuvettes (volume: 51 ml).

Microscale O 2 measurements
O 2 concentration profiles were measured in vertical steps of 100 mm with Clark-type O 2 microelectrodes with a guard cathode [50] (OX-10 Unisense A/S, Aarhus, Denmark) mounted on a manually operated micromanipulator (Mä rtzhä user, Wetzlar, Germany) and connected to a picoammeter (PA2000, Unisense A/S, Aarhus, Denmark) and a strip chart recorder (BD25, Kipp&Zonen, Delft, Netherlands). The microelectrodes had a measuring tip diameter of 10 mm, a stirring sensitivity of ,1-2% and a 90% response time of ,1 second. Linear calibration of the electrode was done from electrode readings in seawater at 100% atmospheric saturation and in anoxic seawater (by addition of sodium dithionite).  Table 2). Prior to these measurements, manually drilled holes (diameter: ,340 mm) on the side of the plastic cuvettes ( Figure 8D, E) were covered with agar (1.5% w/w in seawater) and the cuvettes were left to acclimatize for 2-3 days in order to restore the environment prior to disturbance. The O 2 concentration was measured through the holes of the cuvette wall towards the enclosed bone surface and further into the bone ( Figure 8C). Measurements are given for seven sites on WB1 and five sites on WB2.
The area specific O 2 flux (J, nmol O 2 cm 22 sec 21 ) at the bone surface was calculated from linear parts of the O 2 concentration profiles in the DBL and agar plug according to Fick's first law: Where dC/dz = the change in O 2 concentration (dC, nmol O 2 cm 23 ) with distance (dz, cm), D 0 = the molecular diffusion coefficient of O 2 (cm 2 sec 21 ). We applied a diffusion coefficient of 1.3382 * 10 25 cm 2 sec 21 , corrected for experimental temperature and salinity. Additional microscale O 2 measurements were performed through the mucus tube wall and at the surface of the palps of the individual on WB1. On the mucus tube, the O 2 concentration was measured from outside the tube and half way into the mucus tube wall. Undisturbed measurements could not be obtained within the centre hole of the tube due to contractions of the worm. On the palps, the O 2 concentration was measured by keeping the electrode measuring-tip at a fixed position, while the palps approached the tip position as the annelid protracted after being left undisturbed.