L-Proline Transport Systems of Starfish Pyloric Caeca

Cellular transport of the amino acid L-proline has been extensively investigated in a wide variety of eukaryotic organisms, including yeasts (Horak and Rihova, 1982; Jayakumar et al. 1979), invertebrates (Giordana et al. 1989; Pajor and Wright, 1989; Behnke et al. 1990), fish (Vilella et al. 1988) and mammals (Christensen, 1975; Hayashi et al. 1980; Stevens et al. 1982; Stevens and Wright, 1985). These studies have shown, using inhibitors that selectively block specific carrier mechanisms, that L-proline is transported across most plasma membranes by the combination of Na⁺-dependent and Na⁺-independent processes.

Starfish do not possess an intestine for nutrient absorption as do other types of echinoderms (e.g. sea cucumber, sea urchins), but instead display an array of digestive diverticula of the pyloric stomach called pyloric caeca. These are blind-ended, epithelium-lined tubules believed to be involved in a variety of functions, including the absorption of food after digestion in the stomach (Lawrence, 1982; Lawrence and Lane, 1982; Ferguson, 1979, 1982; Jangoux, 1982). Until recently, the only means of assessing the functional properties of this remarkable organ was indirectly by cytological and histological techniques (Anderson, 1953, 1979; Jangoux and Perpeet, 1972a,EXBIO_158_1_477C23b; Jangoux, 1981, 1982). Recent application of purified epithelial brush-border membrane vesicle (BBMV) techniques to the pyloric caecum of the starfish Pycnopodia helianthoides provided preliminary information suggesting that the cells of this organ possess amino acid transport mechanisms with several properties in common with those of more thoroughly studied mammalian nutrient absorptive epithelial cells (Ahearn, 1990).

The present investigation is a continuation of our earlier preliminary study (Ahearn, 1990) of amino acid transport by starfish pyloric caecum. In this paper, selective inhibitory compounds, effective in characterizing amino acid transport in mammalian cells, are used to define the brush-border carrier components of L-proline transport in starfish cells and to estimate their individual contributions to total transmembrane movements of this amino acid.

Starfish (Pycnopodia helianthoides) were collected from June to August from waters near the Friday Harbor Marine Laboratory of the University of Washington in the San Juan Islands and maintained unfed in flowing sea water until needed. Two to three arms were removed from an individual initially possessing 15–20 arms and the dissected diverticula were pooled for further treatment. The procedure for producing purified brush-border vesicles from this pooled sample was generally the same as the magnesium precipitation technique previously used to make similar membrane preparations from mammalian (Kessler et al. 1978) and crustacean (Ahearn et al. 1985; Behnke et al. 1990) epithelia.

The pooled pyloric caecal sample was homogenized in hypotonic buffer and mixed with 10 mmol l⁻¹ MgCl₂ for selective precipitation of most cellular membranes except the brush border. This was followed with purification by centrifugation at 3000g and 27000g and homogenization in additional hypotonic buffer using a glass homogenizer. A further sequence of precipitation, purification and homogenization in the proper internal buffer was then carried out, followed by centrifugation at 27000 g. The resulting purified sample of brush-border membrane was resuspended in a small volume of internal medium by passage 10–15 times through a syringe fitted with a 22-gauge needle. This final vesicle suspension exhibited a total protein content of approximately 18 mg ml⁻¹ (Bio-Rad proteinassay). Using this preparative method, previous studies with gastrointestinal and renal organs from such diverse organisms as mammals and crustaceans produced final purified vesicle suspensions that exhibited significant enrichments of brush border enzyme markers such as alkaline phosphatase, sucrase and leucine aminopeptidase, while concurrently displaying reduced occurrence of enzymes associated with other cell membranes such as Na⁺/K⁺-ATPase (basolateral membranes) and cytochrome c oxidase (mitochondrial membranes) (Kessler et al. 1978; Ahearn et al. 1985; Behnke et al. 1990).

Transport studies using these pyloric caecal brush-border membrane vesicles (BBMV) were conducted at 18°C using the Millipore filtration technique of Hopfer et al. (1973). At the beginning of a transport experiment a volume (e.g. 20 μl) of membrane vesicles was added to a volume of radiolabelled medium (e.g. 160μl) containing L-[2,3,4,5-³H]proline (New England Nuclear, Corp.). Following incubation periods of 15 s, 1, 2, 5, 10, 20 or 120 min a known volume of this reaction mixture (20μl) was withdrawn and plunged into 2 ml of ice-cold stop solution (composition varying with experiment, see Figure legends). The resulting suspensions were rapidly filtered through Millipore filters (0.45 μm pore diameter) to retain the vesicles and washed with another 5 ml of stop solution. Filters were added to ICN Ecolume scintillation cocktail and counted for radioactivity in a Beckman LS-9000 scintillation counter. Proline uptake was expressed (using the specific activity of amino acid in the external medium) as pmol mg⁻¹ protein filter⁻¹. Each experiment was repeated at least twice using membranes from different animals, yielding qualitatively similar results from each experiment. Within a given experiment each point was analyzed with 3–5 replicates and values are presented in figures as means±S.E.M.

To estimate the effects of external pH on L-proline transport by starfish BBMV, a long-term experiment of the uptake of radiolabelled amino acid was conducted in the presence of an inwardly directed NaCl gradient with internal and external media of pH7.5 and 5.5. Vesicles were loaded with 300mmoll⁻¹ mannitol and either 20 mmol l⁻¹ Hepes-Tris (pH 7.5) or 20 mmol l⁻¹ Mes-Tris (pH 5.5) and were subsequently incubated in media containing 0.1 mmol l⁻¹ L-[³H]proline and 150 mmol l⁻¹ NaCl at either pH 5.5 or pH 7.5. One group of vesicles at each pH was incubated in external medium including 10 mmol l⁻¹ unlabelled L-proline to serve as an inhibitor of carrier-mediated radiolabelled L-proline uptake.

Fig. 1 shows that L-[³H]proline uptake was greater in medium at pH 7.5 than at pH5.5. In addition, uptake at the higher pH led to a transient accumulation of the amino acid to a concentration exceeding that at equilibrium (120min incubation). Only equalizing transport of the amino acid occurred at the lower pH. Addition of 10 mmol l⁻¹ unlabelled L-proline to the external medium at each pH significantly educed the uptake of the amino acid, suggesting that carrier-mediated entry of the organic solute accounted for much of the BBMV uptake observed over the incubation interval. These results suggest that L-proline undergoes concentrative transport at pH 7.5 by at least one carrier process using the transmembrane Na⁺ gradient to transfer the amino acid temporarily against a concentration gradient.

Stevens and Wright (1985) showed that the compound L-pipecolate is a strong competitive inhibitor of the IMINO carrier system for L-proline transport in mammalian intestinal brush-border membrane. To determine the extent to which starfish pyloric caeca transport L-proline by this carrier protein, a similar experiment was conducted in the present investigation. Preliminary time course experiments showed that the uptake of this amino acid by starfish BBMV was a relatively slow process and remained a linear function of time for at least 1 min at both high (5 mmol l⁻¹) and low (0.05 mmol l⁻¹) concentrations. Therefore, for estimation of L-proline influx under a variety of experimental treatments either 1-min or 30-s incubations were used to ensure unidirectional movements of the radiolabelled substrate.

To ascertain the potential inhibitory effect of L-pipecolate on L-proline influx, vesicles were loaded with 300mmoll⁻¹ mannitol, 50mmoll⁻¹ KC1, 20mmoll⁻¹ Hepes-Tris (pH 7.5) and 25 μgml⁻¹ valinomycin (potassium ionophore) and were then incubated for 1 min in external media at the same pH containing 150 mmol l⁻¹ NaCl, 50 mmol l⁻¹ KC1, one of the following concentrations of L-pipecolate: 0, 0.01, 0.025, 0.05, 0.10, 0.25 or 0.50mmoll⁻¹, and either 0.05 or 0.10 mmoll⁻¹ L-[³H]proline. Non-specific isotope binding to vesicles and filters was subtracted from resulting uptake values.

Fig. 2A shows the effect of increasing concentrations of L-pipecolate in the external incubation medium on 0.05 and 0.10 mmol l⁻¹ L-proline influxes in starfish BBMV. This figure indicates that L-pipecolate acted as a potent inhibitor of L-proline entry at each amino acid concentration. Maximal L-proline influx inhibition occurred when vesicles were incubated in medium with 0.5 mmol l⁻¹ L-pipecolate. Even at the highest concentration of the inhibitor, significant amounts of radiolabelled L-proline entered BBMV by processes not inhibited by L-pipecolate. To estimate the contribution of L-pipecolate-sensitive carrier transport to total L-proline influx, it was assumed that at 0.5 mmol l⁻¹ L-pipecolate total inhibition of L-proline influx by the L-pipecolate-sensitive system occurred. L-proline entry at 0.5 mmol l⁻¹ L-pipecolate was subtracted from the amino acid influxes at all other L-pipecolate concentrations and the remainder plotted in the Dixon plot (Fig. 2B). The results of this Dixon analysis indicate that L-pipecolate acted as a potent competitive inhibitor of L-proline influx in starfish BBMV, yielding a K₁ value of 0.02 mmol l⁻¹. From this relationship it is likely that L-proline entry by the L-pipecolate-sensitive process is analogous to transfer by the IMINO system in mammalian epithelia (Stevens and Wright, 1985).

The contribution of different carrier systems to total L-proline influx was determined by incubation of the labelled amino acid in varying concentrations of compounds known to inhibit L-proline transport by the mammalian IMINO, NBB (neutral brush-border) and L (L-leucine preferring) systems (Stevens et al. 1982, 1984; Stevens and Wright, 1985). To make this estimation, starfish BBMV were loaded with 300 mmol l⁻¹ mannitol, 50 mmol l⁻¹ KC1 and 25 μgml⁻¹ valinomycin at pH7.5 and incubated for Imin in media of the same pH containing either 150 mmol l⁻¹ NaCl or 150 mmol l⁻¹ choline chloride, and 0.05 mmol l⁻¹ L-[³H]proline, 50 mmol l⁻¹ KC1 and a series of L-pipecolate, L-leucine and L-alanine concentrations.

Fig. 3 shows the effects of the above inhibitors on the influx of a fixed concentration of L-proline. In the absence of any inhibitors, 0.05 mmol l⁻¹ L-[³H]proline influx was approximately 460 pmol mg⁻¹ protein min⁻¹. Increasing the external L-pipecolate concentration from 0 to 0.5 mmol l⁻¹ reduced L-proline influx to approximately 250 pmol mg⁻¹ protein min⁻¹ as a result of the total inhibition of the IMINO transport system. Addition of L-leucine at concentrations of 0-0.5 mmol l⁻¹ to an external medium already containing 0.5 mmol l⁻¹ L-pipecolate further lowered L-proline influx to approximately 200 pmol mg⁻¹ protein min⁻¹ by blocking L-proline transport by the L system. Adding 1.0mmoll⁻¹ L-alanine to external media containing 0.5 mmol l⁻¹ L-pipecolate and 0.5 mmol l⁻¹ L-leucine led to a further reduction in L-proline influx to approximately 125 pmol mg⁻¹ protein min⁻¹ by eliminating the NBB system. Adding all three inhibitors at maximal concentration to the choline chloride medium did not cause any further reduction in L-proline entry than that observed with the NaCl medium containing all the inhibitors. This suggests that the only L-proline activity associated with BBMV in the choline chloride medium was due to non-specific binding and diffusional entry. Preliminary experiments were conducted to determine the relative selectivity of pipecolate as an inhibitor of the IMINO system by estimating the effects of this compound on the uptake of L-[³H]alanine and L-[³H]leucine by pyloric caecal BBMV. These experiments suggested that pipecolate had no significant inhibitory effect on the uptakes of these amino acids by this tissue, lending support to the use of this compound as a selective inhibitor of the IMINO system alone. As a result of these experiments, the contributions of the IMINO, L and NBB systems to 0.05 mmol l⁻¹ L-proline influx shown on Fig. 3 can be estimated as 64.5%, 15.5% and 20%, respectively.

The kinetics of L-[³H]proline influx by the IMINO, NBB and L systems were determined by measuring the entry of the radiolabelled amino acid as a function of several external concentrations of L-proline in the presence of saturating concentrations of compounds known to abolish entry by two of the three systems. In the first experiment, the kinetics of L-proline influx by the IMINO system was estimated with 0.5 mmol l⁻¹ L-alanine (blocking the NBB system) and 0.5 mmol l⁻¹ L-leucine (blocking the L system) added to the external media. In this experiment vesicles were loaded at pH 7.5 with 300 mmol l⁻¹ mannitol, 50 mmol l⁻¹ KC1 and 25 μgml⁻¹ valinomycin and were incubated for 30 s in media at the same pH containing 150mmoll⁻¹ NaCl, 50mmoll⁻¹ KC1, the two saturating concentrations of the inhibitors and the following concentrations of L-proline: 0.025, 0.05, 0.10, 0.25, 0.50, 0.75, 1.00, 2.50 and 5.0mmoll⁻¹. Non-specific binding of isotope to membranes and filters was subtracted from the total influx prior to estimation of kinetic parameters.

Fig. 4 indicates that L-proline influx under the above conditions was a biphasic function of external [L-proline]. At L-proline concentrations of 1–5mmoll⁻¹ uptake was linearly related to proline concentration, whereas at the lower substrate concentrations the relationship was curvilinear. Over the entire substrate range, influx followed the combination of Michaelis-Menten kinetics plus an apparent diffusion component, as defined by the equation:

where J is total L-proline influx in the presence of saturating concentrations of L-alanine and L-leucine, J_max is maximal L-proline influx by the IMINO system, K_tis the external [L-proline] resulting in half-maximal influx by the IMINO system and P is the apparent diffusional permeability coefficient of vesicles to L-proline. The apparent diffusional permeability coefficient (P) was calculated using linear regression analysis of the slope between 1 and 5 mmoll⁻¹ L-proline. Subtraction of this calculated apparent diffusional influx from total entry values at each external substrate concentration provided an index of amino acid influx by the IMINO carrier alone. These calculated carrier influx estimates are shown in the Hofstee plot (inset) and quantitative values for K_t and J_max were obtained using linear regression analysis of these data. Mean values and their associated standard errors for K_t, J_max and P from this experiment are displayed in the main body of Fig.4 and in Table 1. Similar quantitative values for each constant were obtained in two replicate experiments.

A similar kinetic analysis was performed to estimate the kinetic parameters of L-proline influx by the NBB system. In this instance saturating concentrations of L-pipecolate (0.5 mmol l⁻¹) and L-leucine (0.5 mmol l⁻¹) were added to each external medium containing radiolabelled L-proline. For this analysis vesicles were loaded and incubated in the internal and external media described above, differing only in the inhibitors added so that transport by the NBB system could be independently assessed. Fig. 5 shows that influx under these conditions showed a similar biphasic pattern to that described in Fig. 4 and equation 1. Kinetic parameters for L-proline influx by the NBB system were estimated as in Fig. 4 and the resulting quantitative values for these parameters from a representative experiment are displayed in the body of Fig. 5 and in Table 1. Similar quantitative values for each constant were obtained in a replicate experiment.

The kinetic constants for L-proline influx by the L system were next assessed. In this case 0.5 mmol l⁻¹ L-pipecolate and 0.5 mmol l⁻¹ L-alanine were added to each external L-proline concentration to saturate influx by the IMINO and NBB systems, respectively. Vesicles were prepared and incubated as described above. Results from a representative experiment, shown in Fig. 6, are similar to those presented in Figs 4 and 5, suggesting that L-proline entry by the L system and diffusion follows equation 1. Estimation of the appropriate kinetic constants for this system obtained from this experiment are displayed on the figure and in Table 1. A replicate experiment provided similar quantitative values for each kinetic constant. It is apparent from reviewing Figs 4–6 and Table 1 that the majority of L-proline influx in starfish BBMV occurred by the IMINO system (J_max is three times that of the NBB or L systems), but that the apparent binding affinities (K_t) of the three systems were very similar.

To ascertain the influence of external NaCl concentration on L-proline influx in tarfish BBMV, vesicles were loaded at pH 7.5 with 800 mmol l⁻¹ mannitol, 50 mmol l⁻¹ KC1 and 25 μgml⁻¹ valinomycin and incubated for 1 min in a variety of external media, at the same pH, containing 0.05 mmol l⁻¹ L-[³H]proline, 50 mmol l⁻¹ KC1 and one of the following concentrations of NaCl: 400, 300, 250, 200, 150,100 or 50 mmol l⁻¹. Osmotic pressure was held constant in all media with the addition of appropriate amounts of mannitol. One group of external media contained 0.5 mmol l⁻¹ L-pipecolate to saturate L-proline influx by way of the IMINO system at each external [NaCl], allowing an estimate of the Na⁺-dependency of the NBB system alone, while the other group lacked the inhibitor and represented the effect of external [Na⁺] on L-proline entry by both IMINO and NBB systems. Non-specific isotope binding to membranes and filters, as well as L-proline influx by apparent Na⁺-independent diffusion, were subtracted from uptake data at each experimental condition.

Fig. 7 indicates that the influx of 0.05 mmol l⁻¹ L-proline was a sigmoidal function of external [NaCl] in the presence and absence of the inhibitor L-pipecolate, with each curve approaching saturation at the highest external NaCl concentrations used. The sigmoidal Na⁺-dependency of L-proline influx suggested that the relationship followed the general Hill equation for binding cooperativity:

where J in this case represents total Na⁺-dependent L-proline influx, J_max is total maximal Na⁺-dependent L-proline influx by all Na⁺-dependent systems, is an apparent affinity constant modified to take into account multisite interactions between two or more sodium ions binding simultaneously to all Na⁺-dependent carrier systems, n is the Hill coefficient and represents an index of the number of sodium ions associated with all Na⁺-dependent carrier systems and [Na⁺] is the external sodium concentration.

The present investigation provides evidence for the occurrence of three separate carrier transport mechanisms for L-proline in the brush-border membrane of starfish pyloric caeca. From their responses to selective inhibitors, the three starfish transporters appear to be analogous to the IMINO, NBB and L carriers identified in mammalian intestine (Stevens and Wright, 1985). Like the mammalian gut transporters, the starfish IMINO and NBB carriers are Na⁺-dependent, whereas the L system appears to be Na⁺-independent (Stevens et al. 1984). In mammals, L-proline transport by the IMINO, NBB and L systems accounts for 60, 35 and 5%, respectively, of the total transmembrane transfer of this amino acid (Stevens and Wright, 1985), whereas in starfish 64.5% of L-proline transport takes place by the IMINO system, but the remaining movements are approximately evenly divided between the NBB and L systems (Fig. 3). Lastly, the apparent K_t value of the starfish IMINO system (0.18±0.05mmoll⁻¹, Fig. 4; Table 1) is not significantly different (P>0.05) from that reported for the same transporter in rabbit jejunum (0.23±0.01 mmoll⁻¹; Stevens and Wright, 1985), providing strong support for the conservation of L-proline transport systems across phyla.

In a previous preliminary publication concerning L-proline transport by starfish pyloric caeca, it was concluded that at least two carrier processes for this amino acid were present in the epithelial brush borders of this gut diverticulum. One exhibited a high apparent affinity for its substrate and displayed Na⁺-dependency, while the other had a low apparent amino acid binding affinity and appeared to be unresponsive to this cation (Ahearn, 1990). An apparent K_t of the high-affinity, Na⁺-dependent transporter was reported as 0.22 mmol l⁻¹ L-proline, while the apparent low-affinity process was a linear function of substrate concentration over the range selected. The apparent diffusional permeability of these vesicles to L-proline was calculated to be 300pmolmg⁻¹protein30s⁻¹ (mmoll⁻¹)⁻¹ L-proline. Results of the present investigation with starfish pyloric caeca support and extend these earlier conclusions. By employing L-pipecolate, L-alanine and L-leucine as inhibitors of the IMINO, NBB and Lsystems, respectively, the apparent affinity constant of each process was disclosed. These values were very similar (ranging from 0.13 to 0.21 mmol l⁻¹; Figs 4–6, Table 1) and were close to that reported for the single high-affinity process found in the earlier study. Without the use of inhibitors that selectively blocked the transport contributions from two of the three carrier processes, this earlier study reported an overall apparent affinity constant that represented an average of the values for the three systems. The average P value for apparent diffusion reported in the present study in Figs 4–6 is 543± 147pmolmg⁻¹ protein 30s⁻¹ (mmolI⁻¹)⁻¹ L-proline, which is not very different from that reported for apparent diffusion in the earlier investigation (see above).

The Na⁺-dependency of epithelial amino acid transport has been reported for a wide range of organisms, including mammals (Mircheff et al. 1982; Stevens et al. 1984; Wright and Peerce, 1984), teleosts (Vilella et al. 1988) and invertebrates (Stevens and Preston, 1980; Preston and Stevens, 1982; Gerencser and Stevens, 1989; Pajor and Wright, 1989; Behnke et al. 1990). Among vertebrates, the transport stoichiometry for Na⁺/L-proline cotransport has been reported as being either 1:1 in teleost intestine (Vilella et al. 1988) or 2:1 in rabbit intestine (Wright and Peerce, 1984). Multiple sodium ions (2, 3 or more) have been reported as cotransport substrates in L-proline transport in invertebrate epithelial cells from worm integument (Stevens and Preston, 1980; Preston and Stevens, 1982), mussel gill tissue (Pajor and Wright, 1989) and crustacean antennal gland (Behnke et al. 1990). In none of these studies was the transport stoichiometry of a specific L-proline carrier defined with the use of inhibitors that cleanly abolished the potential transport contribution from two or more simultaneously operating Na⁺-dependent processes. It is, therefore, uncertain in tissues from vertebrates or invertebrates what the specific Na⁺/L-proline cotransport stoichiometries are for the IMINO or NBB systems. Fig. 7 shows, for the first time, the specific Na⁺/r-proline binding stoichiometries of both the IMINO and NBB systems of starfish pyloric caeca. This figure indicates that each Na⁺-dependent carrier process binds two or three sodium ions with each amino acid molecule. However, this experiment does not indicate how many cations are transported for each molecule of L-proline. In this case, as in others pointed out in a recent publication by Gerencser and Stevens (1989), discussing the energetics of Na⁺/amino acid cotransport of invertebrate carrier systems involving the binding of multiple sodium ions for each organic solute molecule, the specific number of cations involved in driving the uphill transport of the cotransported amino acid has not been determined, and therefore the cotransport stoichiometry is unknown. Some associated cations may serve as essential activators of the transport mechanism and may not themselves be transferred across the membrane, while others may accompany the amino acid from one membrane surface to the other. Additional experiments, such as applying the static head method of stoichiometric cotransport analysis developed by Turner and Moran (1982) for mammalian cells and recently successfully used with crustacean epithelia (Ahearn et al. 1990; Behnke et al. 1990; Balón and Ahearn, 1991), need to be applied to each transport system to establish its specific cotransport ratio.

Echinoderms and arthropods are organisms representing the two major divisions of the animal kingdom based on embryological development. The Protostomia, including flatworms, annelids, molluscs and arthropods, exhibit spiral determinate cleavage and the embryonic blastopore gives rise to both the mouth and anus. In contrast, the Deuterostomia are composed of the echinoderms, hemichordates and chordates and exhibit radial indeterminate cleavage, where the mouth does not develop from the embryonic blastopore, but is formed independently. Other characteristics also differentiate these two main lines of animal development. Both major groups of animals have representatives with extensive gastrointestinal diverticula extending from various portions of the gut. The results of this study, and those of the one published earlier (Ahearn, 1990), provide strong evidence for a possible nutrient absorptive role for the starfish pyloric caeca. If the epithelium of this organ is absorptive, the first step of this process would be solute flow from lumen to cytosol across the brush-border membrane, while the next step in transcellular nutrient transfer would be solute exit across the basolateral cell pole. This study and the earlier publication (Ahearn, 1990) characterize the first step in transepithelial L-proline transport by examining the properties of brush border carrier mechanisms. Future experiments defining organic solute transport properties of the pyloric caecal basolateral membrane and studies examining transepithelial nutrient transport by isolated cells from this organ may provide a clear picture of the absorptive nature of this gut diverticulum in echinoderms.

The L-proline transport properties of the epithelial brush border of this structure are very similar to those reported for the same amino acid in the mammalian jejunum (Stevens et al. 1982, 1984; Stevens and Wright, 1985), suggesting that this organ may be the major site of amino acid absorption in this group of echinoderms. This tentative conclusion agrees with results obtained over the last decade concerning the nutrient absorptive role of the crustacean hepatopancreas, another invertebrate organ that is an extensive diverticulum of the pyloric stomach (Ahearn, 1987, 1988; Ahearn and Clay, 1988; Ahearn et al. 1991). In the crustaceans this organ displays a wide range of absorptive and secretory processes for sugars, amino acids and ions. Because representative organisms from the two major divisions of animal development possess gastrointestinal diverticula with similar apparent absorptive capabilities, this function for a little-understood, but widely distributed, structure may be more universal than presently appreciated.

This study was supported by a grant from the National Science Foundation (DCB89-03615). The authors would like to acknowledge and thank Dr A. O. Dennis Willows, Director of the Friday Harbor Laboratories, and the support staff, for providing space and facilities to conduct the work presented in this paper. Specimens of the starfish Pycnopodia helianthoides were provided by Dr Richard Miller.