Cytology reveals diverse cell morphotypes and cellin-cell interactions in normal collector sea urchins Tripneustes gratilla

Echinoderms such as sea urchins are important in marine ecosystems, particularly as grazers, and unhealthy sea urchins can have important ecological implications. For instance, un explained mortalities of Diadema antillarum in the Caribbean were followed by algal overgrowth and subsequent collapse of coral reef ecosystems. Unfortunately, few tools exist to evaluate echinoderm health, making management of mortalities or other health issues problematic. Hematology is often used to assess health in many animal groups, including invertebrates, but is seldom ap plied to echinoderms. We used a standard gravitometric technique to concentrate fixed coelomocytes from the collector sea urchin Tripneustes gratilla onto microscope slides, permitting staining and enumeration. Using Romanowsky stain and electron microscopy to visualize cell details, we found that urchin cells could be partitioned into different morphotypes. Specifically, we enumerated phagocytes, phagocytes with perinuclear cytoplasmic dots, vibratile cells, colorless spherule cells, red spherule cells, and red spherule cells with pink granules. We also saw cell-in-cell interactions characterized by phagocytes apparently phagocytizing mainly the motile cells in cluding red spherule cells, colorless spherule cells, and vibratile cells disproportionate to underlying populations of circulating cells. Cell-in-cell interactions were seen in 71% of sea urchins, but comprised <1% of circulating cells. Finally, about 40% of sea urchins had circulating phagocytes that were apparently phagocytizing spicules. The coelomic fluid collection and slide preparation methods described here are simple, field portable, and might be a useful complementary tool for assessing health of other marine invertebrates, revealing heretofore unknown physiological phenomena in this animal group.


INTRODUCTION
Sea urchins are a keystone species in coral reef ecosystems, because their grazing activities are critical to maintaining balance between faster-growing macroalgae and slower-growing corals (Ogden & Lobel mortality was followed by a mass mortality of dominant Acropora corals leading to a widespread restructuring of the reef ecosystem from highly 3-dimensional arborescent to low encrusting corals, resulting in a marked reduction in diversity of biota (Aronson & Precht 2001). The cause of the Dia dema mortality, thought to be infectious, was never determined.
Loss of grazing fish and sea urchins in Hawaii has resulted in the establishment of invasive algae, a particular problem in Kaneohe Bay, Oahu (Stimson et al. 2001). Since 2014, the State of Hawaii has tried to control invasive marine macroalgae in Kaneohe Bay using a mechanical suction device to aspirate algae from corals (Westbrook et al. 2015). To complement these efforts, the state has also undertaken a rearing program to raise and release as well as translocate collector sea urchins Tripneustes gratilla into the bay (Westbrook et al. 2015). Collector sea urchins were historically present in Kaneohe Bay (Ogden et al. 1989) but have since disappeared for unknown reasons. Reintroduction of sea urchins was done in recognition that this dominant species is an important grazer in Hawaiian coral reef ecosystems (Stimson et al. 2007), with the hope that sea urchins would contribute to long-term control of invasive algae; success of this effort is still being evaluated. Given the importance of T. gratilla to coral reefs, and because translocation and hatchery-release activities can have important health ramifications in terms of disease transfer, we developed a tool to assess antemortem health in this species.
Hematology is commonly used to assess health of various animals (Schalm 1962). For sea urchins, hema to logical exams are usually limited to examining live cells in hanging drops or hemocytometers (Johnson 1969b, Silva 2013. Using these methods, 4 types of coelomocytes are recognized in sea urchins: phagocytes, red spherule cells, colorless spherule cells, and vibratile cells (Chia & Xing 1996, Silva 2013, Smith et al. 2018. Three additional morphological types of phagocytes (discoidal, polygonal, and small phagocytes) have also been described on wet mounts (Smith et al. 2006); however, morphology of a given live cell can vary with time (Liebman 1950, Boolootian & Giese 1958. The phagocytes are the most numerous morphotype and appear to be the first line of host defense by means of phagocytosis. The role of colorless spherule cells and vibratile cells is less clear, but red spherule cells have been shown to be elevated in inflammatory response (Chia & Xing 1996, Silva 2013 or environmental stress (Pin sino et al. 2008). In contrast to examination of live cells, reports of stained cytologic preparations of sea urchin coelomocytes are rare; we could only find 3 reports: Johnson (1969a) used a battery of cytochemical stains including Giemsa to describe coelo mo cyte morphology in Stron gylocen tro tus sp.; Quei roz & Cus to dio (2015) examined cytologic preparations from Eucidaris tribuloides using toluidine blue, trichrome, or hematoxylin and eosin; and Liebman (1950) described cells from Arbacia punctulata using Wright's stains.
While wet mounts of cells allow enumeration of cells ml −1 of coelomic fluid and observation of live cell behavior, examination of Romanowsky-stained slide preparations, a routine part of hematologic exams in other animals, is simple to do and permits a more detailed enumeration of morphologies not readily recognizable in wet mounts of live cells (Schalm 1962). Here, using standard hematology techniques, we identified a variety of morphotypes of coelomocytes in T. gratilla. We also show that cell-in-cell interactions (apparent phagocytosis of sea urchin coelomocytes by sea urchin phagocytes), a heretofore undocumented phenomenon in sea urchins, is common in healthy T. gratilla and is targeted to particular cell morphotypes.

Sample collection, preparation, and light microscopy
In 2016, 20−40 collector sea urchins were sampled each month from 2 sites on South Oahu using SCUBA; these sites were not part of the areas where sea urchins are being released. Sea urchins were held (< 30 min) in seawater in 20 l buckets whilst awaiting coelomic fluid sampling and were released back into the ocean immediately thereafter. Sea urchins were weighed to the nearest 1 g, and test diameter was measured to the nearest 1 mm at the widest point using plastic calipers. Because echinoderm coelomocytes have a propensity to clot or de form rapidly when withdrawn from the coelomic cavity (Liebman 1950, Boolootian 1962, we withdrew 0.05 ml of coelomic fluid directly into 0.05 ml of fixative. The fixative was prepared by mixing 10% formalin with artificial seawater (ASW; Instant Ocean) made according to the manufacturer's instructions. Fixative was preloaded into 1 cc syringes with a 22-gauge, 2.57 cm needle, and coeolomic fluid was drawn by inserting the needle ca. 2−5 mm into the peristomial membrane. In this way, the cells were fixed and killed immediately, ensuring minimal morphologic distortions that often occur in coelomocytes once outside the host (Liebman 1950, Boolootian & Giese 1958. To make the slides, we used a modified gravitational sedimentation system (Moore 2017) comprising a microscope slide, 2 barrels cut from 1 cc syringes, small binder clips, and No. 2 Whatman filter paper cut to fit the microscope slide (Fig. 1). Two holes were punched in the paper over which we centered and affixed the 2 syringe barrels using binder clips. Fixed coelomic fluid (0.05 ml) was dispensed directly onto the slide in each barrel, and the filter paper wicked away fluid, thereby concentrating coelomocytes into a small enough area to allow enumeration without cells piling upon themselves. Once dried, the slides were stained with Romanowsky (Romanowsky 1891) stain (Protocol Hema 3, Ther mo Fisher Scientific) using standard techniques. We found from preliminary empirical trials that this slide preparation/staining method gave superior cell morphology and density rather than the classic push smear or cover slip method (Perpiñan et al. 2006) that proved too damaging to cells and resulted in unrealistically low cell densities to permit efficient cell counting (Liebman 1950). Moreover, the presence of 2 spots of cells allowed the user to choose which gave the best distribution and density for enumeration.
For differential counts, 400 coelomocytes stained with Romanowsky were counted for each sea urchin and enumerated to phagocytes, phagocytes with cytoplasmic perinuclear dot, vibratile cells, colorless spherule cells, red spherule cells, red spherule cells with pink granules, and apparently phagocytized spicules. Phagocytes apparently phagocytizing other cells (cell-in-cell interactions) were enumerated separately. In cases of cell-in-cell interactions, the cell type being apparently phagocytized was also enumerated. Counts of different cell morphologies were converted to percentages. To assess possible seasonal trends, we plotted size distribution of sea urchins and percent cell morphotypes by month.

Percoll gradients and separation of cells
To relate morphology of coelomocytes seen on Romanowsky stain to that seen in live cells, we used Percoll density centrifugation to segregate cells into smaller groups. Live coelomocytes were sampled from Tripneustes gratilla by withdrawing 4 ml of co elo mic fluid directly into 4 ml of ice-cold (4°C) anticoagulant buffer (ACB) formulated exactly as described by Gross et al. (2000). Percoll was made isotonic with ASW by blending 9 parts of Percoll to 1 part of 10× ASW. Isotonic Percoll was then layered in a 15 ml conical tube in 2 ml layers comprising 90% Percoll overlaid by 75, 65, and 45% layers for a total of 8 ml. Four ml of coelomic fluid in ice-cold ACB was then carefully overlaid on the Percoll gradient, tubes were centrifuged at 1500 × g (30 min at 4°C) in a swing rotor centrifuge, each layer subsequently harvested, and rinsed twice in 10 ml of cold ASW. Fractions were observed as wet mounts, used to make Romanowsky-stained slides as above, or used for electron microscopy (Section 2.3). Because Romanowsky staining sometimes degranulated red spherule cells, these were confirmed morphologically by locating the fully granulated cells on unstained slides with a Lovins Micro-Slide Field Finder and relocating the same cells after staining.

Electron microscopy
For electron microscopy, 1 ml of coelomic fluid was withdrawn into 1 ml of modified Karnovsky's fixative (Karnovsky 1965) comprising 2% paraformaldehyde and 2.5% glutaraldehyde in ASW; the cells were Fig. 1. Gravitational sedimentation system used to make coelo mic fluid slides for Tripneustes gratilla. Microscope slide, filter paper with holes, clips, and 1 cc syringe barrels cut to size (A) prior to and (B) after assembly. Ruler is in cm then centrifuged at 1000 × g for 1 min, fluid was decanted, and cells were overlaid with fixative. These cells and Percoll layers (Section 2.2) were processed as described by Work et al. (2017). Briefly, cells were rinsed in sodium caco dy late buffer with 0.45 M sucrose, post fixed with 1% osmium tetroxide, dehydrated in ethanol and propylene oxide, embedded in resin, sectioned, stained with uranyl acetate/lead citrate, and viewed on a Hitachi HT7700 transmission electron microscope (TEM) at 100 kV. Identity of cells was confirmed by repeating this process using cells fractionated from Percoll layers (Section 2.2) fixed for electron micro scopy as phagocytes, vibratile cells, colorless sphe rule cells, and mixed red spherule cells and red spherule cells with pink granules.

Statistics
Test diameter and percent cell counts between months were compared with non-parametric Kruskal-Wallis ANOVA (Kruskal & Wallis 1952), because data did not fit assumptions of normality as determined by the Shapiro-Wilk test (Shapiro & Wilk 1965). Because these 9 comparisons were not independent, we did a Bonferroni adjustment of the significant p-value (0.05/9 or 0.005) to account for the number of sta tis tical comparisons (Rice 1989). Tukey pairwise post hoc comparisons were made in cases where differences were significant. All analyses were done with R version 3.5.3 (R Development Core Team 2011).

RESULTS
We sampled 340 sea urchins from January− December 2016. The mean ± SD (range) test dia -meter and weight were 73 ± 11 mm (31−103 mm) and 155 ± 67 g (18−388 g). Test diameter and percent differential counts did not differ significantly between months.
Phagocytes were the most common cell type, comprising a mean of 73% of cells and characterized by a homogeneous amorphous pale blue cytoplasm (Table 1). Live phagocytes ranged from petalloid to fillipodial to amoeboid depending on time (5−60 min) under observation. On Romanowsky, phagocytes with cytoplasmic perinuclear dot comprised about 8% of the cell population and were distinguished from regular phagocytes by the presence of a distinct deeply basophilic perinuclear cytoplasmic dot in the cytoplasm ( Fig. 2A,B). Colorless spherule cells comprised about 5% of cells. Live cells were round to oblong with densely packed, variably sized intracyto plasmic granules (Fig. 2C) and moved slowly by extending cytoplasm that then filled with granules (see Video S1 at www. int-res. com/ articles/ suppl/ d142 p063_ supp/). On Roma now sky, intracytoplasmic granules of colorless spherule cells stained deep blue (Fig. 2D). Rapidly erratic and non-directional motile live vibratile cells with characteristic flagella (Fig.  2E) comprised about 7% of cells. On Romanowsky, these cells had a pink cytoplasm with small arrayed granules and a flagellum often detached (Fig. 2F). Live red spherule cells were oblong, had red granules (Fig. 2G), a similar motility as colorless spherule cells (Video S2), and comprised about 4% of cells. When fixed and stained, red spherule cells typically degranulated and were packed with clear vacuoles (Fig. 2H). The degranulation phenomenon during fixation/ staining in red sphe rule cells was confirmed by locating specific cells with a Lovins Micro-Slide Field Finder prior to and after staining when red granules were and were not visible, respectively. Red spherule cells with pink granules comprised <1% of cells, and on wet mounts were similar to but smaller than colorless spherule cells with small granules. On Romanowsky, pink granule cells had densely packed pink intracyto plasmic granules (Fig. 2I). Ap parently phagocytized spicules comprised <1% of cells and were notable by the presence of variably sized small to large bi hamate crystalline structure spanned by a delicate mesh of cytoplasm with an eccentric nucleus ( Fig. 2J− L). More than 98% of sea urchins sampled had phagocytes, red and colorless spherule cells, and vibratile cells, over half had red spherule cells with pink granule cells, and a third had apparently phagocytized spicules (Table 1). Percoll layers isolated 4 populations of cells. The top layer (45% Percoll) had phagocytes and phagocytes with intracytoplasmic perinuclear dots; the second layer (65% Percoll) had colorless spherule cells; the third layer (75%) had vibratile cells; and red spherule cells and red spherule cells with pink granules went to the bottom of the tube. Apparently phago cytized spicules were not seen on Percoll assays. This segregation of cell populations with Percoll allowed us to identify particular cell types on electron microscopy and relate them back to light micro graphs. On ultrastructure, phagocytes had numerous intracytoplasmic vacuoles (Fig. 3A) whilst phagocytes with a perinuclear cytoplasmic dot had a variably sized reticulated to electrondense punctate structure near the nucleus (Fig. 3B). Colorless sphe rule cells (Fig. 3C) were closely packed with membrane-bound granules with electron-lucent cores. Red spherule cells had numerous electron-lucent gra nules with electron-dense cores (Fig. 3D). Vibra tile cells were distinguished by presence of a single flagellum with numerous intracytoplasmic granules, some of which had an electron-dense core (Fig. 3E). Due to their scarcity, we were unable to obtain TEM images of apparently phagocytized spicules or red spherule cells with pink granules.

DISCUSSION
The use of a gravitational sedimentation system in combination with immediate fixation of coelomocytes permits preparations of cells suitable for examination on cytology and is a low-cost, simple technique to do hematology exams for Tripneustes gra tilla that is likely applicable to most echinoderms. This method offers several advantages over examination of live cells on phase contrast (Johnson 1969b), including (1) storing of samples for later examination, making this technique amenable for remote field situations; (2) increased details of cell morphology as the result of Romanowsky staining; and (3) visualization of phenomena (cell-in-cell interactions) previously undocumented in echinoderms. As such, this method could complement existing techniques that depend on examination of live cells to provide data on numbers of coelomocytes ml −1 of coelomic fluid, examine changes in cell morphology over time, and evaluate hematological responses of sea urchins to various environmental changes (Matranga et al. 2005, Pinsino et al. 2008. The percentages of vibratile cells, red and colorless spherule cells, and phagocytes (Chien et al. 1970) in T. gratilla are within the ranges of those documented for other species of echinoderms. For instance, in his re view of echinoderm immunology, Silva (2013) found the following ranges: phagocytic phagocytes, 60− 70%; vibratile cells, 15−20%; red spherule cells, 5−10%; and colorless spherule cells, 5−10%. In comparison, we found mean percentages of 73, 7, 4, and 5% for phagocytes, vibratile, red, and colorless spherule cells, respectively; these ranges certainly fit within parameters outlined in the literature (Table 1). Moreover, Romanowsky-stained slides revealed 2 additional morphologies, i.e. phagocytes with a cytoplasmic perinuclear dot and red spherule cells with pink granules. The nature of the perinuclear cytoplasmic dot in type 1 phagocytes could not be determined from electron microscopy; however, we saw no evidence on TEM that they were infectious organisms like bacteria or virus inclusions (Cheville 1976), so further research is needed to determine their origin. Red spherule cells with pink granules sedimented at the same layer as red spherule cells on Percoll sedimentation and on live preparation look like smaller versions of colorless spherule cells with smaller granules. However, unlike the red spherule cells, red spherule cells with pink granules did not degranulate with fixation, suggesting a different composition of their intracytoplasmic granules. Apparently phagocytized spicules with their biha-mate crystalline structure are reminiscent of spicules seen in tissues of gonads and tube feet of Echinometra sp. (Kelso 1970, Rahman et al. 2001). However, the presence of bihamate spicules among circulating coelomocytes has rarely been documented in echinoderms, the only instance being in Echinometra sp. from the Pacific where they were transiently abundant in circulation (Boolootian 1962). We cannot confirm whether the cell structures (nuclei and diapha nous membrane spanning the spicule) were phago cytes attempting to phagocytize wayward spicules that were sloughed from tissues or actual cells engaged in spicule morphogenesis. The presence of incomplete bihamate spicules associated with these cells (Fig. 2L) suggests to us the latter as a possibility; moreover, in sea urchins, spicules originate from cells (Horstadius 1939), so this is a reasonable conjecture. The scarcity of this cell type makes sorting out their origin a challenge, but perhaps one avenue might be to look at other echinoderm species, such as Echino metra sp., where apparently phagocytized spicules are reportedly more numerous (Boolootian 1962). Given that red spherule cells with pink granules and apparently phagocytized spicules comprise <1% of the cell population and were found in 54 and 34% of sampled sea urchins, respectively, it is not surprising that they have not been documented previously. Determining whether similar proportions of these cell types are found in other sea urchins might be a fruitful avenue of research.
The ultrastructure of the 4 major Tripneustes coelomocytes is similar to those for Strongylocentrotus sp. (Chien et al. 1970, Vethamany & Fung 1972. We suspect that Vethamany & Fung (1972) might have misidentified colorless and red spherule cells be cause colorless spherule cells in their study looked more like red spherule cells in Tripneustes (this study) and Strongylocentrotus (Chien et al. 1970). Possibly these cell types were mis-identified on ultrastructure for Strongylocentrotus, because in that study (Vethamany & Fung 1972), cells were not segregated using sediment gradients prior to TEM.
The phenomenon of cell-in-cell interactions seen in 71% of sea urchins sampled here indicates that this phenomenon must have some biological importance. That phagocytes are enveloping target cells is evident because of the presence of continuous plasma membrane of phagocytes around apparently phagocytized cells as seen on electron microscopy ( Fig. 3F−  L). Interestingly, the cells within the phagocytes appeared intact with no evident visible patho logy, and al though it might be tempting to conclude that phagocytes were phagocytizing cells, we cannot state this definitely. Cell-in-cell interactions vary depending on the mechanisms involved (Brown et al. 2015, Fais & Overholtzer 2018. In classical phagocytosis of cells, phagocytized cells are usually apoptotic (Chao et al. 2011); however, on light and electron microscopy, we saw none of the morphologic hallmarks of apoptosis, such as membrane blebbing, karyor rhexis, or apoptotic bodies (Elmore 2007), nor did we see evidence of necrosis such as karyorrhectic nuclei (Cheville 1976) in cells that were apparently phagocytized by phagocytes.
Phagoptosis is a process in mammals, worms, and flies whereby phagocytes ingest intact cells based on presentation of specific membrane markers that signal to phagocytes whether or not the target cells should be ingested (Brown & Neher 2012). Phagoptosis leads to death of the ingested cell; however, we could not determine the fate of ingested cells in this study nor did we document the presence of plasma markers that serve as signals for phagocytes to phagocytize cells.
Entosis occurs when one cell actively invades the cytoplasm of another cell, leading to death of the invading cell (Overholtzer et al. 2007), and is found most often in cancers. The physics of entosis stipulate that the invading cell has stiffer cytoskeletal structures than the one being invaded (Ning et al. 2015), and entosis involves specific membrane markers and presence of adherens junctions between the cell membranes, something not seen on TEM here. Emperipolesis (Humble et al. 1956) is a more general term describing cell-in-cell interactions involving different cell types that usually lead to death of invading or invaded cells (Xia et al. 2008). In the case of T. gratilla, given that we do not know the fate of either the phagocytes or the cells within them, we have opted to be more conservative and term what we saw here a cell-in-cell interaction. Sorting out the precise mechanisms of cell-in-cell interactions in T. gratilla will likely require development of membrane markers and documenting the fate of the interacting cells, which may be challenging given that cell-incell interactions comprise <1% of cell populations.
Phagocytosis in echinoderms is well documented (Bertheussen & Seljelid 1978, Bertheussen 1981a; however, most such studies have looked at phagocytosis of materials such as latex beads, bacteria, or India ink iatrogenically introduced into sea urchins. We know of only one instance of cell-in-cell interactions of red spherule cells in coelomocytes of Strongylocentrotus after 7 h of observation in vitro (Chien et al. 1970). Whilst the cell-in-cell interactions seen here might be an artefact of processing or could have occurred after placement on slides, we think this unlikely because cells were fixed immediately in formalin upon withdrawal from the sea urchin. Indeed, fixing coelomocytes in sea urchins was used over 70 yr ago to ensure consistent cell morphology for microscopic examination (Liebman 1950).
We are unaware of other documentations of cell-incell interactions in the coelomic fluid of live, apparently healthy echinoderms. Interestingly, the prevalence of cell-in-cell interactions did not reflect the underlying relative numbers of circulating cells (Table 1). For instance, red spherule cells comprised ca. 4% of circulating coelomocytes but 55% of apparently phagocytized cells. Possibly, red spherule cells are preferentially targeted because of their propensity to degranulate and release echinochrome, a potent iron-scavenging compound, into the coelomic fluid (Coates et al. 2018), which in large amounts might be detrimental to the host. Colorless spherule and red spherule cells with pink granules were also preferentially targeted by cell-in-cell interactions; however, given the uncertain function of these cells, this is difficult to explain. In contrast to the aforementioned cells, cell-in-cell interactions with phagocytes was relatively rare (comprising < 3% of instances; Table 1) despite the fact that this is the most abundant cell. Perhaps cell-in-cell interactions in sea urchins play a housekeeping role, removing cells from circulation that could be potentially damaging to the host or that have outlived their usefulness. For instance, phagocytosis of neutrophils and red cells in mammals serves to remove senescent cells (Brown et al. 2015), and perhaps a similar mechanism occurs in sea urchins. It seems noteworthy that cell-in-cell interactions predominantly target the motile cells of T. gratilla, perhaps lending some credence to the hypothesis that phagocytes might be actively invaded by motile cells. Cell-in-cell interactions seem to merit further investigation to see if it this phenomenon is widespread in other echinoderms, why exactly certain cell types are targeted over others, and determine its role in regulating numbers of coelomocytes in sea urchins.
Seasonal variation in cell counts in echinoderms has not been investigated; however, it is known that at least for cell function, seasonal variation in physiological function of marine invertebrates does exist. For instance, in shore crabs Carcinus maenas in England, phagocyte activity was higher in winter and spring, whereas coelomic fluid oxidant status was lower in winter (Dissanayake et al. 2011). Here we showed that within limits of the variation in counts, there was no evident seasonality in differential cell counts for T. gratilla in Hawaii. Shifts in percentages of coelomocytes on hematology exams of T. gratilla could be used as an indicator of health. For example,  Pinsino et al. (2008) and Pinsino & Matranga (2015) used a combination of heat shock proteins and coelomocyte counts to evaluate responses of sea urchins to environmental changes, and perhaps the methods outlined here might be used to complement such approaches to assess echinoderm health.