Spiculous skeleton formation in the freshwater sponge Ephydatia fluviatilis under hypergravity conditions

Gemmule harvesting and preparation

Gemmules were collected in the winter months (November 2013–March 2014), from specimens of E. fluviatilis kept in outdoor aquaria at the University of Amsterdam. Sponge tissue containing the gemmules was collected and cleaned by rubbing the sponge tissue between two pieces of corduroy to free the gemmules from the sponge tissue scaffold that can inhibit germination. Detached gemmules were sterilized in 1% H₂O₂ for 5 min on a shaker at 4 °C to remove bacterial and fungal contaminants included in the coat (Funayama et al., 2005).

Gemmule hatching

A total of 420 gemmules from one individual sponge were plated on sterile 12-well culture plates in sterile M-Medium (Rasmont, 1961; one mM CaCl₂, 0.5 mM MgSO₄, 0.5 mM NaHCO₃, 0.05 mM KCl, 0.25 mM Na₂SiO₃) under ambient conditions. The plates were sealed by parafilm. Each well of the 12-well culture plates contained one gemmule in approximately four mL of M-medium. After 4 days, before placing the gemmules in the LDC, the individuals that did not hatch were discarded and of the remaining hatched specimens, only the sponges in stage II were selected (i.e., specimens without any erected spicules or osculum, see Fig. 1A for a description of developmental stages). A total of 72% of the plated gemmules hatched (302 out of 420), of which n = 277 sponges developed to stage II that were subsequently used for the experiment in the LDC (Table 1). Before placing the selected gemmules in the LDC, the medium was refreshed to prevent accumulation of waste products in the well plates.

Sponge feeding with ¹³C- and ¹⁵N-enriched amino acids

For each gravitational condition (see below), one group (i.e., approximately half) of the gemmules was fed with isotopically-enriched (¹³C and ¹⁵N) dissolved amino acids in order to study nutrient uptake, whereas the other group remained unfed (Table 1). The fed groups (n = 22, 20, 25, 29, 26, respectively, for 1, 2.5, 5, 10, and 20 × g) were randomly selected and received M-Medium with 390 μg L⁻¹ tracer ¹³C- (11 μmol L⁻¹) and ¹⁵N- (2.2 μmol L⁻¹) enriched dissolved amino-acids added prior to the LDC runs. The non-fed group (n = 24, 21, 21, 19, 20, respectively, for 1, 2.5, 5, 10, and 20 × g) did was kept in sterile M-Medium throughout the experiment.

Hypergravity experiment in large diameter centrifuge

Hypergravity experiments were performed using the LDC of the ESA (Noordwijk, The Netherlands) following a predetermined protocol (Fig. 2A). The LDC has a diameter of eight m and comprehends four large arms fitted with outward swinging gondolas. The rotational movement of the arms creates an artificial acceleration field at the well plates positioned inside the gondolas, simulating different g-forces depending on gondola placement on the arms and rotational speed (Fig. 2B). Experiments where performed in two 48-h runs. In the first 48-h run developing sponges were exposed to 1 (i.e., as hypergravity control; placed at the center of the rotating LDC), 2.5 and 5 × g and in the second run to 10 and 20 × g (Fig. 2B). Per hypergravity level, both fed and non-fed sponges were tested simultaneously. To assess the effect of rotation on the development of the gemmules, a control experiment without feeding was performed where gemmules were hatched on a non-rotating lab bench, in addition to the rotating control at 1 × g. As conditions did not vary in the LDC room between the first (2.5 and 5 × g) and the second run (10 and 20 × g) no additional hypergravity control at 1 × g was performed during the second run.

Sponge surface area measurements

Before and after placement in the LDC, the surface area of the substrate covered by the sponges (in mm²) was determined by light microscopy. We are aware that sponge volume is a better metric for growth of sponges building a three-dimensional structure (i.e., the skeleton). However, we were unable to accurately measure volume and therefore used the increase of substrate area covered by the basal epithelium of the sponge relative to the substrate area covered by the basal epithelium of the sponge before placement in the LDC, as a proxy for sponge size/growth. All sponges were photographed and visually checked to assess substrate area cover, the set-up/no set-up of the sponge skeleton, and the presence/absence of an osculum under a stereoscopic microscope (Olympus SZH-ILLD with infinity1 microscopy camera). Surface area measurements were performed using Image J software (https://imagej.nih.gov/ij/).

Stable isotope analysis of sponges

After LDC exposure and microscopy imaging, the labeled M-medium was replaced with non-labeled M-medium for 3 × 5 min to remove as much residual label as possible. After the last replacement the sponges were left for 30 min in the non-labeled medium to remove any residual label from the sponge surface. Subsequently, sponges were then taken out of the well-plates, rinsed with M-medium and pooled per treatment, freeze-dried, homogenized, and stored at −20 °C in silver boats. Pooling was necessary to ensure that sufficient carbon and nitrogen was available for the stable isotope analysis. Samples were put in silver boats, acidified with 5% HCl to ensure removal of inorganic carbon, oven-dried at 60 °C, pinched closed and stored frozen before analysis on an Elemental Analyser (EA, Thermo Scientific, Waltham, MA, United States, Flash EA 1112 Analyzer) that was coupled to a Delta V isotope ratio mass spectrometer (IRMS) for simultaneous measurement of ¹³C:¹²C and ¹⁵N:¹⁴N ratios. Reproducibility for the EA-IRMS analysis was 0.25‰ for ¹⁵N and 0.2‰ for ¹³C.

The uptake of dissolved amino acid C or N was expressed as uptake of μmol tracer C or N per mmol sponge C or N per d. Rates are calculated from the delta notations obtained from the IRMS as δX (‰) = (R_sample/R_ref − 1) × 1,000, in which X is the element (C or N), R_sample is the heavy:light isotope ratio in the sample and R_ref is the heavy:light isotope ratio in the reference material (R_ref = 0.0111797 for C and R_ref = 0.0036765 for N). The atomic fraction of the heavy isotope (F) in a sample is calculated as F = R_sample/(R_sample + 1). The excess (above background) atomic fraction is the difference between the F in an experimental sample and the atomic fraction in a control (i.e., non-enriched) sample: E = F_sample − F_control. The excess incorporation of ¹³C and ¹⁵N was multiplied by 1,000 to express rates in μmol tracer C or N per mmol sponge C or N and divided by the incubation time to convert to daily rates.

Statistical analysis

To analyze the effect of rotation without hypergravity exposure on the sponges, a two sample t-test was performed to compare the means of the two (rotating/non-rotating) 1 × g groups. A Chi-square test was used to compare osculum development between the two groups. Both tests were performed in SPSS version 25 (Released 2011, IBM SPSS Statistics for Windows, Version 25.0; IBM Corp., Armonk, NY, USA).

Surface area cover increase is expressed as a percentage relative to the original surface are covered by each developing sponge hatched from a gemmule in stage II at the start of the LDC experiment. The effect of hypergravity on percentage surface increase was investigated by a linear model, adopting a 0.05 significance level. Besides evaluation of the overall model significance, also the pairwise differences among the treatment-levels were made through a Tukey HSD test with a 95% family-wise confidence level.

Osculum formation was analyzed with a generalized linear model using a binomial error function. To address the effect of osculum formation and g-force on the mean dissolved organic matter (DOM) uptake (measured by ¹³C and ¹⁵N enrichment), an additive linear model was evaluated, using a 0.05 significance level.

In the results, means and proportions are reported in combination with the standard deviation between brackets. All analyses were carried out in R statistics programming environment R 3.3.2 (The R Foundation for Statistical Computing, 2004–2013).

Sponge development under hypergravity exposure

The non-rotating and rotating controls at 1 × g did not show a significant difference in sponge growth (two-sample t-test, t(31) = −0.341, p > 0.1), skeleton formation (in both groups 100% of the sponge specimens showed set up skeleton) or osculum development (Chi-square test, χ² (2, n = 33) = 1.8397, p > 0.1).

Over all levels of hypergravity exposure, 96% (±1%) of stage II juvenile sponges developed to stage III juvenile sponges (i.e., with erected skeleton), 66% (±7%) of all sponges reached stage IV (i.e., build an osculum), and formed a fully functional sponge during their 48-h treatment (Table 1). All sponges showed an increase in surface area of on average 196% (±70%) compared with the surface area before placement in the LDC. However, the increase in surface area significantly decreased at higher hypergravity levels (R² = 0.07, n = 227, p < 0.01) (Figs. 3A and 4). Comparing the different treatment levels (using a Tukey HSD test), only the relative surface area change between 1 vs. 20 × g (−47%, p < 0.01) and 5 vs. 20 × g (−45%, p < 0.01) were significant.

No effect of feeding on surface area increase was found (Fig. 4). A significant interaction effect of g-force and feeding on mean osculum formation per group was observed, where fed sponges formed oscula less frequently than non-fed sponges, especially at higher g-force (logit (osculum formation) = 0.71−0.13 feeding + 0.06 g-force − 0.10 feeding g-force; Nagelkerke-R² = 0.08, n = 227, p < 0.05).

Tracer isotope incorporation

Amino acid assimilation was confirmed by significant enrichment in both δ¹³C (42.5 ± 11.1‰) and δ¹⁵N (1,202.8 ± 280.2‰) in the tissues of all fed sponges compared with background, non-fed, sponge tissue (−30.5‰ δ¹³C and 18.2‰ δ¹⁵N). On average, hatching E. fluviatilis assimilated 0.4 ± 0.1 μmol C_{amino acids} mmol C⁻¹_sponge d⁻¹ and 4.3 ± 1.0 μmol N_{amino acids} mmol N⁻¹_sponge d⁻¹ (Fig. 5). In total, 9.1% ± 3.7% and 41.2% ± 18.5% of the added amino acid carbon and nitrogen, respectively, was processed by the sponges during the 48-h incubations. No significant effect of g-force was found on ¹³C-assimilation rates, while the overall effect for ¹⁵N assimilation rates was significant (C∼g-force, n = 10, p = 0.165; N∼g-force, n = 10, p < 0.05) (Fig. 5). When comparing the effects on ¹⁵N for different treatment levels (using a Tukey HSD test), only the difference between 5 × g vs. 20 × g (−2.7, p < 0.01) appeared significant.

Spiculous skeleton construction and osculum formation under hypergravity

Sponges are known for their enormous plasticity and opportunistic nature when it comes to food selection (Ribes, Coma & Gili, 1999; McMurray et al., 2016) and regeneration after environmental stress (Ayling, 1983; Wulff, 2010; Alexander et al., 2015). This is particularly important in ecosystems with large seasonal fluctuations in water temperature and food availability, where the formation of gemmules by freshwater sponges is a very successful strategy to survive these adverse and possibly lethal conditions. When conditions are more favorable, gemmules hatch and develop into fully grown sponges in order to enter a successful sexual reproductive stage and obtain sufficient biomass to gemmulate (Funayama et al., 2005; Manconi & Pronzato, 2016). The formation of the spiculous skeleton is a crucial feature of the morphogenesis of freshwater sponges, resulting eventually in the formation of an active filter-feeding system (Rozenfeld, 1980; Funayama et al., 2005), enabling the sponge to grow and reproduce. Albeit the hypergravity forces used in our experiment are not naturally occurring, this is the first in vivo study of freshwater sponges to show that skeleton formation of E. fluviatilis is highly resilient under high levels of stress. Even after the juvenile sponges were exposed to hypergravity forces up to 20 × g for 48 h, they managed to survive, grow, create and organize their spicules, and develop a functional food-uptake system. In comparison, a well-trained human astronaut has a g-force tolerance of approximately 9 × g for only short periods of seconds to minutes (Wu et al., 2012). Our initial hypothesis that increased g-forces decrease the ability of the sponge cells to transport and erect their spicules and prevent the formation of a fully-developed filter system (e.g., no osculum formation) is clearly rejected. Sponges showed a significantly smaller increase of substrate area covered at higher g-force, which may suggest impaired growth, but almost all specimens (96 ± 1%) were able to set-up their spiculous skeleton and the majority was able to create and osculum (66 ± 7%).

It is noteworthy that, although this study focused on the siliceous spicules of the E. fluviatilis skeleton, the fibrous matrix of spongin and possibly chitin in which the spicules are embedded, may also play a role in withstanding environmental stress (Ehrlich et al., 2010, 2013). The presence of chitin is known to be responsible for the rigidification or skeletal structures in invertebrates including marine demosponges (Brunner et al., 2009; Ehrlich et al., 2010) and has since long been known to be an important component of E. fluviatilis gemmules (Zykoff, 1892). However, the presence of chitin has not been reported in adult specimens of E. fluviatilis and it is at present unknown whether or how the fibrous matrix of the skeleton changes in response to environmental conditions.

Fed vs. non-fed sponges under hypergravity

We also show that hatching E. fluviatilis can take up dissolved nutrients (i.e., amino acids). Previous evidence corroborates that freshwater sponges can feed on dissolved organic food sources, such as dissolved proteins (Weissenfels, 1976; Manconi & Pronzato, 2008). However, this study shows, for the first time, that developing sponges hatched from gemmules take up and incorporate dissolved food sources in their tissue possibly even before arranging their filter system and start active pumping. In that case, the dissolved ¹³C- and ¹⁵N-labeled amino acids can be acquired either via passive diffusion or through phagocytosis by choanocytes or surface pinacocytes (Willenz & Van De Vyver, 1982; Willenz, 1980). However, the uptake of dissolved amino acids did not affect skeleton formation nor surface area growth. We therefore conclude that the added nutrients are not a necessity for the developing sponges to overcome the environmental stress caused by hypergravity. Since most of the sponges developed normally even under extreme levels of stress, this suggests that the process of erecting the spiculous skeleton relies mainly on intrinsic energy contained within the gemmule. Interestingly, fed sponges developed oscula significantly less frequently compared with non-fed sponges, especially at exposure to the highest (10 and 20 ×) g-forces. We speculate that under extremely stressful conditions, when energy costs are highest, sponges invest in the formation of an active filtration system to acquire new energy. Additional food can help prolong the process of hatching and skeleton development, which might increase their chances of survival throughout severe stress conditions.

The role of dissolved food in the diet of freshwater sponges

The capacity of developing E. fluviatilis to readily take up dissolved food sources could be a valuable addition to the daily natural diet of adult stages of the sponge. Evidence is accumulating that many sponge species utilize DOM as major part (71–94%) of their daily carbon intake (Yahel et al., 2003; De Goeij et al., 2008a; Mueller et al., 2014; McMurray et al., 2016), however, these studies were all done on marine sponge species.

The uptake of DOM is hypothesized to be influenced by the presence of associated microbes. It has been shown that mainly sponges with high numbers of associated bacteria (high microbial abundance sponges) utilize DOM as food source, as opposed to the low microbial abundance (LMA) sponges (Hoer et al., 2018), unless sponges have an encrusting growth form (De Goeij, Lesser & Pawlik, 2017). The freshwater sponge E. fluviatilis is a LMA species, but the algal symbionts of E. fluviatilis are not an obligate requirement for survival of the host (Wilkinson, 1980). Moreover, E. fluviatilis can be both encrusting and massive, depending on its stage of development. In this study, we cannot conclude on the role of associated microbes in the processing of DOM. Although all gemmules were sterilized before the experiment, the presence of associated microbes in the sponge specimens cannot be excluded. Moreover, by determining bulk assimilation of the isotope tracer we cannot distinguish whether DOM was processed by symbionts or occurred directly be sponge cells. Future studies using bacterial-and sponge-specific fatty acid biomarkers (De Goeij et al., 2008b, Rix et al., 2017) may identify these processes. It would be interesting to assess whether E. fluviatilis adjusts its diet depending on its algal symbionts and/or at various phases of its life-cycle.

In tropical coral reefs, sponges play a key role in the cycling of nutrients by converting predominantly DOM into particulate organic matter through high cell turnover and detritus production, which is subsequently shunted to higher trophic levels—a process termed the sponge loop (De Goeij et al., 2013). Since detritus production and rapid division of choanocytes by mitosis have been reported for E. fluviatilis (Weissenfels, 1976; Tanaka & Watanabe, 1984), similar sponge-loop processes potentially also occur in freshwater systems. Additionally, the C:N ratio of the amino acids (5:1) in the medium was significantly lower than the assimilated C:N ratio by E. fluviatilis (1:5), and the sponge assimilated up to four times more nitrogen from the amino acids (41%) as compared with amino acid carbon (9%). Carbon could have been partly lost to respiration, but E. fluviatilis most likely assimilated nitrogen selectively from the amino acid source. As sponge detritus in eutrophic tropical ecosystems was found relatively enriched in nitrogen compared to sponge tissue (De Goeij et al., 2013), and the current view is that limitation of nitrogen is also common in eutrophic freshwaters (Downing & McCauley, 1992), freshwater sponges might fertilize their surrounding ecosystems in the same way as their marine counterparts.