Apical Membrane Permeability of Mytilus Gill: Influence of Ultrastructure, Salinity and Competitive Inhibitors on Amino Acid Fluxes

Normal ciliary activity of the lamellibranch gill results in a flow of water through the mantle cavity that is associated with gas exchange, acquisition of particulate foodstuffs and the removal of wastes from the animal. During this activity, the gill is continuously exposed to sea water which has much lower concentrations of small organic solutes than the cytoplasm. Among these compounds are the amino acids which are used by bivalve tissues in isosmotic volume regulation (Gilles, 1979). Intracellular concentrations of amino acids in the gill are greater than 100 mmol l⁻¹ cell water (Zurburg & De Zwaan, 1981; Wright & Secomb, 1986), whereas the concentration of free amino acids in the near-shore environment ranges from 20 nmol l⁻¹ to 2 μmol l⁻¹ (Manahan & Arnold, 1983; Braven, Evens & Butler, 1984; Siebers & Winkler, 1984). These extremes of concentration are separated by the apical membrane of the gill epithelium, and movements of solute across this membrane may play a role in cell volume regulation (Crowe, 1981) and in the acquisition of nutrients (Wright, 1982).

The permeability of the apical membrane is influenced, at least in part, by the presence of several separate, carrier-mediated pathways for amino acid transport (Wright, 1985). However, though the gill can effect a net accumulation of amino acid from dilute solution (Manahan, Wright, Stephens & Rice, 1982; Manahan, Wright & Stephens, 1983; Siebers & Winkler, 1984; Wright & Secomb, 1986), the very large intracellular concentration of these substrates has led to the suggestion that the passive loss of amino acids from integumental tissues may be large (Hammen, 1968). It has been proposed that apical transport processes cause a reduction in the loss of endogenous substrates from the integument (Gomme, 1981), and recent evidence has supported this suggestion in marine bivalves (Wright & Secomb, 1986). It has also been suggested that such processes could be influenced by the presence of the microvillous brush border of the apical membrane (Gomme, 1981), though this has not been examined critically.

Flux of amino acids across the membranes of bivalve integumental tissues has been shown to be influenced by several factors. Reduction of ambient salinity has been shown to decrease rates of amino acid uptake into isolated gill tissue (Anderson, 1975; Anderson & Bedford, 1973; Bamford & Campbell, 1975) and to increase rates of efflux from intact animals (Livingston, Widdows & Fieth, 1979) and isolated tissues (Henry & Mangum, 1980). However, it is not clear whether such factors influence general membrane permeability, or preferentially influence the flux of separate classes of these compounds. A specific loss of taurine from integumental tissues of Mytilus edulis can be produced by acute exposure to submillimolar concentrations of structural analogues of taurine (Wright & Secomb, 1986).

In the present study we have examined the structure of the apical membrane of gill cells, and have considered the relationship between membrane structure and the kinetics of integumental transport using a mathematical model. The results indicate that the microvillous brush border membrane of gill cells does not influence the kinetics of transport in these cells, though it is clear that an appropriate combination of morphological and physiological parameters could result in a significant role for the brush border in influencing solute exchange. We have also examined the effect on amino acid efflux from the gill of (i) reduced salinity and (ii) exposure to high concentrations of specific structural classes of amino acid. The former procedure resulted in a transient increase in the general apical membrane permeability to amino acid, while the latter treatment increased the permeability to specific structural classes of these compounds.

Mussels (M. edulis and M. californianus) were purchased from the Bodega Bay Marine Biological Laboratory, and were kept in a refrigerated (13°C) aquarium containing filtered artificial sea water (Instant Ocean). Encrusting organisms were scraped from the valves with a wire brush. The animals were not fed and were usually used within 4 weeks of collection.

Experiments in which intact animals were exposed to reduced salinity were performed as follows. Valves were propped open by insertion of a 0·7 mm glass tube (i.d. 0·2 mm) into the mantle cavity ventral to the exhalant aperture. Large rubber bands secured the valves such that the tubes remained in place. By means of a small plastic cannula (P.E.90), fed through the glass tube, and a peristaltic pump, the mantle cavity could be continuously irrigated with the test medium in which the animal was placed. Pump-driven flow was approximately 25 ml min⁻¹ for a 10-to 30-g animal (wet mass soft-body parts). The incubation volume was usually 500 ml, and all test solutions were based on the artificial sea water (ASW) described by Cavanaugh (1956). Efflux of amino acid from animal tissue was measured by sampling the medium and measuring the rate of appearance of total amino acid as determined using the chromatographic procedure described below. Control experiments were performed to verify that efflux of amino acid was dependent upon exposure of the medium to the mantle cavity; incubations of the test chamber or of animals completely sealed with rubber bands had no effect on amino acid levels in the test solution.

Studies of the effect of exogenous application of amino acid on amino acid efflux used a protocol similar to that described above, except it was not necessary to prop the animals open; the animals were permitted to gape and pump normally.

For studies on isolated gill tissue, gills were removed from specimens of M. californianus and a piece of 000 silk suture tied to one end of each demibranch. The tissue was held in ASW for 30 min prior to any experimental incubation. Experiments were started by suspending a demibranch in a test solution (usually 50 ml) that was gently mixed with a magnetic stirrer. Test solutions, and the medium used for the last 5 min of the preincubation, contained l0 mol l⁻¹ 5-hydroxytryptamine (5-HT) to activate lateral cilia (see Wright, 1979). Fluxes of amino acid were determined as described for experiments using intact animals.

Accumulation of ¹⁴C-labelled amino acids was measured using procedures described previously (Wright, 1985). Isolated demibranchs from M. californianus were suspended in 200 ml of ASW of appropriate composition, containing 10 μmol l⁻¹ 5-HT and 0·5 μmol l⁻¹ labelled amino acid. After 5 min, the tissue was removed and rinsed in ice-cold ASW for 5 min. Disks (7 mm) of tissue were cut from the rinsed tissue, and accumulated radioactivity extracted in 0·7ml of 0·1 mol l⁻¹ HNO₃. After 2h, 10 ml of scintillation cocktail was added and radioactivity measured in a scintillation counter (Beckman 3801). All counts were corrected for quench using H-number analysis.

High performance liquid chromatography (HPLC) was used to measure amino acids in samples of ASW. The procedure was a modification of that described by Lindroth & Mopper (1979) and Manahan et al. (1983). The reagent o-phthal-dialdehyde (OPA) was used to make fluorescent derivatives of the amino acids in experimental samples. The derivatives were separated on a 3 gm C18 reverse-phase column (Beckman ODS Ultrasphere) and monitored with a fluorescence detector. Individual peaks were identified and quantified by comparison of retention times and peak areas to those of appropriate standards. The reproducibility of standard retention times and peak areas was better than 5%. The limit of detection and quantification for taurine was 0·7 pmol, which corresponded to a concentration in the test solutions of 4 nmol l⁻¹. The limit for alanine, aspartate, glutamate, glycine and serine was approximately 1 nmol l⁻¹.

Specimens of M. edulis were dissected in sea water. Gill tissue was removed intact and prepared for fixation for transmission electron microscopy. Tissues were initially ‘quick fixed’ for 20s (Bradley & Satir, 1981) in a 1:1 mixture of 2% osmium tetroxide and 14 mol I^-1 sodium cacodylate + 0·6 mol l⁻¹ sucrose. Tissues were then rinsed in buffer and placed for 1 h in a buffered glutaraldehyde solution containing 4 % glutaraldehyde, 0·05 mol l⁻¹ cacodylate buffer (pH 7·3) and 0·3 mol l⁻¹ sucrose. Tissue was postfixed in 1 % osmium for 1 h in the above cacodylate buffer and dehydrated in an ethanol gradient and propylene oxide. While in propylene oxide, smaller pieces of gill were removed from the fixed intact tissue and then embedded in Epon 812.

Thin sections were taken perpendicular to the long axis of the gill filaments. Randomly selected thin sections containing circumferential cross-sections of gill filaments were sequentially photographed at ×6600, recording the entire circumference of the gill filaments. Using a Zeiss digital analyser MOP 3 at a final magnification of ×16 500, linear measurements were made of the gill surface membrane, excluding the surface area represented by the microvilli. Measurements were then made of the microvillar surface area, i.e. the highly folded apical membrane itself. These data were separated into three categories, each representing a distinct region of the cross-section, i.e. frontal, sublateral and abfrontal. Data for each region, collected from three filaments, were combined and analysed statistically.

Analysis of the percentage of gill surface area occupied by intramicrovillar and extramicrovillar space was conducted on the micrographs using a linear analysis on the Zeiss MOP 3 at the level of the glycocalyx. Estimations of intermicrovillar distances were made by measuring the distance between adjacent microvillar profiles at the level of the glycocalyx.

OPA was purchased from Cal-Biochem. [¹⁴C]alanine (150 mCi mmol⁻¹) was purchased from ICN-Radiochemicals. [ⁱ⁴C]taurine (97·5 mCi mmol⁻¹) was purchased from New England Nuclear. All other chemicals were purchased from standard sources and were the highest grade available.

The gills of Mytilus produce a unidirectional flow of water through the mantle cavity. The general architecture of the gill and the pattern of water movement through this tissue are described in several excellent discussions (e.g. White, 1937; Bayne, Thompson & Widdows, 1976; Jørgensen, 1976). For the present purpose it is sufficient to emphasize that all the water that passes through the animal must pass through the ostial network of the gill, during which time this water briefly (<0·2s) comes within approximately 20 μm of the external surface of gill cells. The lamellar surfaces of the demibranchs through which the inhalant water passes consist of numerous parallel gill filaments which are attached to their neighbours via junctions, either ciliary (in the case of M. edulis) or fleshy (in the case of M. californianus).

Each gill filament is a tube whose outer surface is exposed to sea water while the inside is bathed by haemolymph. In the subsequent discussion, each filament is subdivided into three distinct regions: Frontal, Sublateral and Abfrontal (Fig. 1). Transmission electronmicrographs of filaments from M. edulis revealed that each region consists of three distinct layers (Figs 2, 3) : (i) an outer layer formed by a sheet of epithelial cells which sits on (ii) a fibrous lamina of collagen-like material (the conchiolin of White, 1937) and (iii) an inner layer of thin endodermal cells that appear to provide an incomplete lining of the branchial blood vessel.

The epithelial cell layer, composed of the cells facing sea water, represented the predominant cellular mass of the gill. The Frontal region was composed of: three ciliated cell types, named on the basis of the type of cilia borne by each, i.e. frontal, laterofrontal and lateral; goblet cells; and columnar epithelial cells. The Abfrontal region included a ciliated cell type, as well as goblet and columnar cells. The cells of the sublateral region (the largest percentage of total gill surface) became progressively squamous in form (Figs 2, 3). There were no ciliated cells in this region, but goblet cells were noted.

Microvilli were a structural element of the apical membranes of all the filament surfaces (Figs 2, 3). Morphometric analyses were performed on the ‘brush border’ of the three regions of the gill (Fig. 1): (i) frontal, i.e. the region including the lateral ciliated cells and the tissue above them; (ii) abfrontal, i.e. the region of the ciliated cells of the abfrontal aspect of the filaments; and (iii) sublateral, i.e. the area of the sides of the filament below the lateral ciliated cells and above the first ciliated cells of the abfrontal region. No significant differences were found to exist in the diameter and spacing of microvilli in the three regions. However, the Sublateral microvilli were significantly shorter (P< 0·05) than those of the Frontal and Abfrontal aspects of the filaments (0·6 vs 1·2 and 1·1 μm, respectively). This explains, at least in part, why the microvilli of the Sublateral region caused a comparatively smaller increase in apical cell area than the Frontal and Abfrontal microvilli, i.e. 10·3-fold vs 18·5-and 16·5-fold, respectively. Nevertheless, given the length of this perimeter, the Sublateral aspect of the filaments still provided 46% of the gill surface area (Table 1). We previously estimated that there are approximately 550 cm² of cell apex area per gram wet mass of mussel gill tissue (approximately equal for both M. edulis and M. californianus; Wright & Secomb, 1984). The present morphometric measurements indicate that total apical area due to microvilli is actually of the order of 7300 cm² g⁻¹ gill tissue.

There was a glycocalyx associated with the microvilli of all three regions, and it was particularly evident towards the tips of the microvilli (Fig. 3A). Fig. 3B shows a tangential section through the apex of several cells bearing laterofrontal cilia. The condensation of glycocalyx is evident as the section moves towards the tips of the microvilli in this region. At this level of the brush border, microvillar area represented approximately 40 % of the total of extracellular area plus microvillar area (Table 1).

Taurine constituted between 57 and 86 % of the free amino acid pool of the gill and mantle from M. edulis and M. californianus (Table 2). Taurine, in conjunction with aspartate, glutamate, serine, glycine and alanine, made up in excess of 95 % of the non-ammonia, OPA-sensitive material that was resolved chromatographically. The total, extractable pool of amino acid from the gill was 79//mol g ‘wetmass for Al. edulis and 133 μmol g⁻¹ for M. californianus.

When intact specimens of Al. edulis and Al. californianus were placed into aminoacid-free ASW there was a gradual increase in taurine concentration of the medium which reached a steady state generally within 30–60 min. The steady-state concentrations for taurine averaged 17 ± 8·6 nmol (S.E., N = 4) and 26 ± 5·7 nmol.(N = 6) for Al. edulis and Al. californianus, respectively, in close agreement with previous measurements (Wright & Secomb, 1986).

The measured rate of loss in these experiments of the ‘non-taurine’ constituents of the integumental amino acid pool was more equivocal. The limit of resolution of our chromatographic procedures permitted quantification of amino acids other than taurine at concentrations as low as 1 nmol l⁻¹. However, this level is within the range of the background contamination of our system for amino acids such as serine, glycine, alanine, aspartate and glutamate. It also approximated the steady-state concentration for these substrates. For example, in seven experiments with Al. californianus, there was no significant difference between the concentrations of the non-taurine amino acids in the zero-time ASW blank and the concentration of these substrates after 60min. Taurine concentration, on the other hand, was significantly greater after 60 min (P=0·02; Wilcoxon matched-pairs signed-ranks test). Nevertheless, our observations do permit assigning an upper level to the steady-state concentration for the non-taurine amino acids (i.e. [amino acid]_ss in Al. californianus (N=7; in nmol l⁻¹): [Asp]_ss< 6·9; [Glu]_ss<4·2; [Ser]_ss< 12·1; [Gly]_ss < 13·4; [Ala]_ss<5·6. Upper limits for steady-state concentrations for M. edulis were (N = 4): [Asp]_ss< 13·8; [Glu]_ss<3·6; [Ser]_ss< 17·6; [Gly]_ss < 16·5 ; [Ala]_ss<9·5.

When intact M. edulis were exposed to a step decrease in salinity, from 32‰ to 19·2‰ (i.e. from 100 % to 60 % ASW), there was an immediate increase in the rate of loss of amino acid from the animals (Fig. 4). For example, during a 15-min exposure to 60% ASW taurine loss increased more than 44-fold; the average loss of taurine went from the control value of 0·52 ± 0·179 nmol g⁻¹ wet mass soft-body parts to 23·4 ± 7·69 nmol g⁻¹. The result of this increase in loss of taurine from mussels was the development of ambient taurine concentrations that ranged from 0·2 to 3·0 mol l⁻¹(N = 4). Expressed per gram of gill tissue, the change in taurine efflux was from 6·3 ± 1·13 (S.E., N=4) to 333 ± 131·6 nmolg⁻¹15 min⁻¹. Though this latter set of figures is likely to overestimate the actual loss of taurine from the gill, the gill does represent approximately 85 % of the total surface area of the mantle cavity (Jørgensen, 1983), and normalizing efflux data to gill mass can be expected to provide a reasonable approximation of the effect of an acute exposure to a hypotonic medium on taurine loss from the gill.

Exposure to 60 % ASW also resulted in a loss of the non-taurine constituents of the integumental free amino acid pool (Fig. 4). In four experiments with M. edulis, the exposure to reduced salinity increased efflux (compared to control values) by: Asp, 10·1-fold; Ala, 6·5-fold; Gly, 7·3-fold; Glu, 10·3-fold; and Ser, 2·5-fold. Identifiable peaks of Tau, Asp, Glu, Ser, Gly, Ala and NH₃ accounted for 79·8 ± 2·95% (N = 4) of the OPA-positive material on the chromatograms of the samples collected after the 15-min exposure to 60% ASW.

The increased efflux of amino acid resulting from exposure to reduced salinity was at least partially reversible. Following the exposure to 60% ASW, the mussels were exposed to 100 % ASW for 1 h, after which the medium was changed and loss to the medium determined after 15 min (Fig. 4). Though in most cases the ‘recovery’ rates of efflux were higher than the control rates (approximately one-to three-fold), they represented a 41–92% decrease in the maximal efflux noted during exposure to reduced salinity. Therefore, the loss of amino acid induced by exposure to hypotonic medium was not due to irreversible damage to the apical surface of integumental cells.

Exposure to a hypo-osmotic shock resulted in a similar, reversible increase in amino acid efflux from intact M. californianus. In two experiments, loss of taurine was increased 88-fold over the control rate [15·1 ±8·90 (the range of two experiments) to 1330 ± 179 nmolg⁻¹ 15 min⁻¹], and recovery was 95% complete (i.e. recovery period efflux of 76·4 ± 18·20 nmolg⁻¹ 15 min⁻¹). Effluxes of the other major constituents were as follows (in nmolg⁻¹ 15 min⁻¹, pre-test to 60% ASW to recovery): Asp, 2·7 to 108·4 to 7·6; Glu, 2·9 to 57·0 to 2·7; Ser, 4·1 to 46·8 to 6·0; Gly, 4·3 to 146·0 to 12·4; Ala, 1·4 to 176·8 to 9·4.

Gill tissue isolated from M. californianus lost amino acids to 100% ASW at rates higher than those noted in studies with intact mussels. For example, the apparent efflux of taurine during the first 5min (5·8 ± 0·65 × 10⁻⁷ mol g⁻¹ h⁻¹; N = 6) was approximately 10 times higher than the steady-state loss of taurine from gills in intact mussels (Wright & Secomb, 1986). This difference was probably due to the presence of damaged tissue at the cut edges of the in vitro preparation. The non-taurine amino acids were also lost from gill tissue, although at 2–5 % of the rate observed for taurine, which was consistent with their,contribution to the total free amino acid pool in gill tissue.

Isolated gills could be given a controlled exposure to the experimental solutions. Isolated gills treated with 5-HT retained lateral ciliary activity so that all gill surfaces were exposed to the test solution.

When gills from M. californianus were exposed for 5 min to a step decrease in salinity (100 % to 60 % ASW) there was a rapid, reversible increase in the rate of loss of all the major pool amino acids (Fig. 5). The increase in efflux caused by exposure of the gill to 60% ASW was greatest for taurine: a 36-fold increase (from a control rate of 48·2 ±5·39 to 1760 ± 238 nmolg⁻¹ 5 min⁻¹). For the other major amino acids, in decreasing order, the increases in efflux were : Asp, 27·7 ± 6·9; Gly, 13·9 ± 4·0; Glu, 9·8 ± 1·9; Ala, 9·4 ± l·7; Ser, 3·4 ± 0·5. Two experiments in which gill tissue was exposed to 60% ASW for 15 min indicated that the rate of net loss of these amino acids was not a linear function of time for more than 5 min, but appeared to approach a steady state after approximately 10–15 min (data not shown). Thus our estimates of the rate of amino acid loss induced by exposure to 60 % ASW (Fig. 5) may underestimate the initial efflux across the apical membrane of gill cells. The 5-min loss of these amino acids represented a maximum of 2·3% (for taurine) and a minimum of 0-9 % (for glutamate) of the total pool of each in the gill.

There was a progressive decline in the amount of amino acid lost from isolated gill tissue when it was exposed to 60% ASW for five successive 5-min washes. In the experiment shown in Fig. 6, the initial exposure to 60% ASW resulted in a 36-fold increase (compared to the control period) in the rate of taurine efflux from the tissue. This was reduced to a 12-fold increase during the fifth wash. The non-taurine amino acids showed a similar progressive reduction in the rate of efflux induced by exposure to the hypo-osmotic medium. Less than 4 % of the total tissue taurine was lost during the five rinses (3·5 μmolg⁻¹), suggesting that the decline in taurine efflux noted in Fig. 6 was not due to a simple depletion of this compound from the tissue.

Exposure to a hypo-osmotic medium also inhibited the influx of amino acid into isolated gill tissue. Accumulation of 0·5 μmol l⁻¹ [¹⁴C]alanine during a 5-min exposure to 60 % ASW was reduced by 79 %, compared to the control condition in 100 % ASW (0·35 ± 0·06 μmolg⁻¹h⁻¹vs 1·73 ± 0·36 μmolg⁻¹ h⁻¹ ; N = 4). Uptake of 0·5 μmol l⁻¹ [¹⁴C]taurine was reduced 82 % by the same treatment (0·19 ± 0·08 vs 1·01 ± 0·27 μmolg⁻¹ h⁻¹ ; N= 5). Consequently, the increased net loss of amino acid from gill tissue to 60% ASW must have involved, at least in part, the reduced capacity of several, separate influx pathways to reaccumulate this material (see Wright & Secomb, 1986). The degree to which reduced ‘recycling’ of substrates influenced the observed loss of amino acids was presumably a function of the rate of loss of each compound and the kinetics of its uptake. In the case of taurine, the 36-fold increase in net loss caused by exposure to 60% ASW led to an efflux that exceeded by two-fold the maximal rate of influx noted in a previous study (Wright, 1985). Such a rate of loss cannot easily be ascribed to an inhibition of taurine uptake. However, the 10-fold increase in alanine loss corresponds to a net efflux (0·18 gmol g⁻¹ h⁻¹) that was less than 1 % of the maximal rate of influx of this amino acid (≈40 μmolg⁻¹ h⁻¹ ; Wright, 1985). Therefore, inhibition of alanine uptake could have contributed significantly to its loss in 60 % ASW.

We previously found that addition of a specific competitive inhibitor of taurine uptake (y-aminobutyric acid; GABA) resulted in an increase in the rate of efflux of taurine from integumental tissue, while having no significant effect on the loss of other amino acids (Wright & Secomb, 1986). It was concluded that the increase in taurine efflux was the combined result of (i) the competitive inhibition of the reaccumulation of taurine lost from integumental tissues by normal, steady-state processes, and (ii) the accelerative heteroexchange of endogeneous taurine for GABA. Thus it was of interest to examine the effect of adding to test solutions substrates which have been shown to inhibit the transport of other structural classes of amino acid. When intact mussels were exposed to 40 μmol l⁻¹ L-leucine there was an immediate increase in the rate of loss of other neutral (i.e. zwitterionic) α-amino acids, in particular alanine, glycine and serine (Fig. 7B). This increase was transient; after rising to between 40 and 190 nmol l⁻¹, the concentration of these neutral a-amino acids decreased to less than 10 nmol I^-1 after 2h. Over the same period, integumental uptake of the ambient leucine caused its concentration to decrease from 38 μmol l⁻¹ to 21nmol l⁻¹ (Fig. 7A). Of particular importance was the observation that only neutral α-amino acids were lost when leucine was added; there was no appreciable change in loss of taurine (a neutral β-amino acid; Fig. 7C), or of aspartate or glutamate (anionic α-amino acids; Fig. 7D), compared to the steadystate concentrations of each of these substrates. These latter structural classes of amino acid are transported by separate pathways (Wright, 1985), and therefore the failure of leucine to affect the efflux of these amino acids was expected.

The cationic α-amino acid lysine is a partial inhibitor of neutral a-amino acid uptake (Wright, 1985). In two experiments with M. californianus, addition of 50 μmol l⁻¹ lysine resulted in a small, transient increase in the concentration of alanine and glycine (alanine, maximum ambient concentration of 55 nmol l⁻¹vs [Ala]_ss of 4nmol l⁻¹; glycine, maximum of 49 nmol l⁻¹vs [Gly]_ss of 8·3 nmol l⁻¹), with comparatively little effect on the concentrations of serine, taurine, aspartate and glutamate (data not shown). Finally, the addition of 50 μmol l⁻¹ aspartate had no effect on the efflux of either neutral α-or β-amino acids (the presence of the large aspartate peak in chromatograms from these experiments prevented observations of any effects of aspartate on the flux of glutamate; data not shown).

The integumental epithelium of the bivalve gill is routinely exposed to surrounding sea water. As a result, the apical membrane of gill cells serves as a permeability barrier, separating sea water from cytoplasm and, in conjunction with the basal-lateral membrane of these cells, the branchial haemolymph. The present study has examined several aspects of apical membrane permeability to amino acids.

The microvillous brush border of the apical membrane of gill cells increases the area exposed to sea water, consistent with the idea that these cells may have an absorptive function. The permeability of this barrier, as shown by the rate of loss of amino acid to the surrounding medium, is very low and influenced by at least two experimental manipulations: (i) reduction in ambient salinity results in a general increase in the permeability to amino acids; and (ii) exposure to large, external concentrations of neutral amino acid causes a specific increase in the loss of endogenous amino acids.

It is a general axiom that the microvillous brush border of epithelial cells serves to increase the absorptive area of the cells’ apical surface (Fawcett, 1962). Morphological measurements of the microvilli of M. edulis gill verified that the brush border amplifies the total apical area exposed to sea water by 10-to 18-fold (Table 1). This is considerably less, however, than the effect on surface area noted for microvilli from several other systems. For example, the brush border of cells in the mammalian proximal convoluted tubule increases the luminal area by a factor of up to 36 (Welling &’Welling, 1975), and the brush border of the rat intestine increases apical area by a factor of 24 (Palay & Karlin, 1959). As discussed below, the comparatively small area of microvillous membrane may assist in reducing the overall permeability of the apical surface, thereby reducing the diffusional loss of endogenous substrates to surrounding sea water.

Though the presence of microvilli clearly results in an amplification of cell membrane area, it is less clear what effect the geometry of microvilli has on the kinetics of solute exchange across the membrane. It is conceivable, for example, that the combination of microvillous structure and the kinetic characteristics of transporters in the apical membrane could serve to reduce significantly the amount of endogenous material lost from the cell to the overlying medium (see also Gomme, 1981). To examine this possibility, we have developed a theoretical model for the effect of microvillous structure on reaccumulation of amino acids by gill cells. It incorporates the geometric characteristics of the apical membrane and the kinetics of membrane transport (see Appendix for the development of the model). For the purposes of this discussion, we have considered the following hypothetical situation: (i) uniform distribution of the active transport process along the sides of the microvilli and (ii) localization of the ‘leak’ to the basal area of the apical membrane, between the microvilli (see Fig. 8A). Solute lost across the apical membrane must ‘run the gauntlet’ between the microvilli, during which time it can serve as substrate for the transporters and thereby be recovered. This particular configuration maximizes the opportunity for the microvillous brush border to reduce the amount of endogenous material lost from gill cells to surrounding sea water. Results from the theoretical model will therefore provide an upper limit on this effect for a given set of transport and morphological parameters. The parameters associated with the calculations are (i) the ratio of J_max to K_t, i.e. the maximal rate of carrier-mediated transport, expressed per unit area of microvillous membrane, to the Michaelis constant for the transporter^*; (ii) the rate of loss of amino acid, expressed per unit area of the base of the microvilli; (iii) the height of the individual microvilli; and (iv) the perimeter and area density of the microvilli overlying the apical base of the gill cell.

Details of the theoretical model are given in the Appendix. Fig. 8B depicts the computed variation of the rate at which solute diffuses along the spaces between microvilli from the base (z = 0) to the tips of the microvilli (z = z_m). The flux at z = z_m represents the apparent leakage rate from the brush border. Comparison of the three curves shows that this leakage depends strongly on the relationship between the linear (i.e. first-order) uptake rate and the diffusive rate, as represented by the parameter λz_m defined in the Appendix, which depends on the geometric and transport parameters listed above. The value of λz,_n calculated for the apical brush border of M. edulis gill is 0·26 (see Appendix). According to the model, approximately 97 % of the solute leaked at z = 0 emerges from the brush border without being reaccumulated under these conditions. The assumption that leakage occurs only at the base maximizes the opportunity for solute reabsorption. Therefore we conclude that the combination of the kinetics of uptake, configuration of microvilli, and rate of solute diffusion are such that the geometrical configuration of the microvilli has little effect on the apparent leakage of solute from the cells. Once in bulk solution, however, taurine of endogenous origin can be reaccumulated. The efficiency of this reaccumulation is such that approximately 30 % of the material lost from the integumental surface can be recovered (Wright & Secomb, 1986).

It is of interest to note that, just as the structure of the apical membrane appears to have no direct effect on the kinetics of solute efflux, it can likewise be shown that it should not influence the influx of solute. This observation is of interest because previously we estimated the kinetics of amino acid uptake into the pumping gill using a model that took into account the influence on uptake of gill structure and water flow: the ‘convection—diffusion’ model (Wright & Secomb, 1984; Wright, 1985). However, the model did not take into account the effect of microvilli on the kinetics of transport. The present results suggest that the simple convection-diffusion model is adequate to describe transport in the gill, and indicate that our estimates for the Michaelis constants for amino acid uptake into this organ (i.e. 1–5 gmol l⁻ ) provide reasonable estimates of the ‘true’ value for the K_t of each of these membrane processes.

It is worthwhile to emphasize that, though the dimensions of the microvilli in the bivalve gill are such that they would appear to have little impact on the kinetics of solute exchange, this need not be the case for microvilli in other systems. Microvilli can exceed 4·5 μm in length, and the brush border can have a fractional density of 80% (Smith et al. 1969; see also Berridge & Oschman, 1972). For example, if the gill brush border had such dimensions, the presence of microvilli would result in recovery of approximately 32% of the taurine lost from gill. Thus, the structure of microvilli can play a role in defining the effective kinetics of solute exchange between an epithelial cell and the overlying medium, given an appropriate set of morphological and kinetic parameters.

The low steady-state concentrations for the non-taurine amino acids suggest that the permeability of the apical membrane to these compounds is very low. At steady state, the rate of efflux from the animal is balanced by the rate of influx via integumental pathways. The kinetics of uptake for glycine into intact, actively pumping M. californianus have been determined: the J_max is 19·2 × 10⁻¹²mol cm⁻²s⁻¹, with K_t of 4·9gmol l⁻¹ (Wright, 1985)^*. The ambient concentration at which glycine efflux is equivalent to influx in M. californianus is ⩽6·8 nmol l⁻¹. Efflux should then be ⩽ [(19·2 × 10⁻¹²) (6·8 × 10⁻⁹)]/[(4·9 × 10⁻⁶) + (6·8 × 10⁻⁹)] or ⩽2·8 × 10⁻¹⁴molcm⁻²s⁻¹. Given the tissue content of glycine (Table 2), the intracellular concentration of this compound in gill cells is probably at least 2 mmol l⁻¹. Assuming that this approximates the intracellular activity of this compound, the steady-state efflux of glycine reflects an upper limit of the apical membrane permeability to this molecule of 1·4 × 10⁻⁸cms⁻¹. A similar calculation for the permeability of M. californianus gill to aspartate results in an estimated permeability coefficient of 0·6 × l0⁻⁹cms⁻¹. These values are similar to our previously reported estimate of l × 10⁻⁹cms⁻¹ for the permeability of M. edulis gill to taurine (Wright & Secomb, 1986). The actual permeability of the apical brush border will be approximately 13-fold lower than these figures (i.e. 10·8–0·5 × 10⁻¹⁰cms⁻¹), as they do not take into account the effect of microvilli on increasing apical surface area. Finally, to the extent that the efflux of these amino acids is the result of backflux through mediated pathways or paracellular pathways, flux across the epithelial surfaces of the mantle, or excretion, the actual passive permeability of the apical surface of the gills will be lower than our calculations suggest.

It is worthwhile to consider how these estimates for gill membrane permeability compare to estimates from other epithelial systems. Unfortunately, though a number of studies have addressed the question of ‘transepithelial’ membrane permeability to amino acids (e.g. Dantzler & Silbernagl, 1976; Barfuss & Schafer, 1979), there have been few studies examining the passive permeability of the apical membrane of epithelial cells to these substrates. However, studies with isolated brush borders can be used to estimate passive permeability coefficients. Uptakes into isolated membranes are usually expressed per milligram of membrane protein. Permeability coefficients (P) from such studies, therefore, have units of 1 mg⁻¹s⁻¹ (e.g. Stevens, Ross & Wright, 1982). For purposes of comparison, these permeabilities must be converted into units of cms⁻¹. This calculation can be made assuming that the average diameter of a brush border vesicle is 0·2 μm (e.g. Haase, Schafer, Murer & Kinne, 1978; Hopfer, Crowe & Tandler, 1983). Combined with the measured value of the ‘equilibrium’ volume of the brush border vesicle suspension (where 1 μl of ‘apparent space’mg⁻¹ membrane protein = l × 10⁻⁹cm³mg⁻¹), the conversion of P values to units of cms⁻¹ becomes apparent. For example, Stevens et al. (1982) reported P values of 1·6 and 4·5 × l0⁻⁸lmg^-1s⁻¹ for proline and phenylalanine, respectively, in brush border membranes from rabbit jejunum. The equilibrium volume of their vesicle preparation was 1 μl mg⁻¹, resulting in estimated P values of 5 and 15 × l0⁻⁸cms⁻¹ for proline and phenylalanine, respectively. These values are 50–3000 times greater than the conservative values for P in the bivalve gill estimated in the present study. The apparent passive permeability of the gill approaches the values reported for amino acids in liposomal membranes (i.e. ≈1 × 10⁻¹¹ cms⁻¹ ; Young & Ellory, 1977). This very low permeability of gill membranes, in conjunction with a comparatively sparse apical brush border, may in part account for the ability of gill cells to sustain such large intracellular amino acid concentrations in the face of the constant exposure of the apical membrane to sea water.

Exposure to an abrupt decrease in ambient salinity resulted in a dramatic increase in the rate of loss of amino acid from gill tissue, and served to illustrate that gill cells contain amino acids which are osmotically active rather than bound or compartmentalized. The present set of observations cannot distinguish between specific, carrier-mediated efflux of amino acids and a general increase in non-specific membrane permeability to these compounds. However, the reversibility of the process (Figs 4, 5) suggests that the exposure to a hypo-osmotic medium did not result in irreversible damage to the apical membrane of the gill epithelium. Furthermore, the transient nature of the loss of amino acid from M. californianus gill tissue (Fig. 6) suggests that membrane permeability may be regulated. It should also be stressed that the similarity of the responses to reduced salinity of M. edulis and M. californianus was expected. While the euryhalinity of M. edulis is widely documented (e.g. Lange, 1963), the capacity of M. californianus to adapt to reduced salinity is not always appreciated. Though largely confined to the comparatively stenohaline environment of the open, rocky coast, this restriction appears to be the result of the stenohalinity of the larvae of M. californianus ; adult M. californianus survive for many months at salinities of 20 (Young, 1941).

There are several other reports of an increase in loss of amino acids from bivalves upon exposure to dilute sea water. For example, Livingston et al. (1979) reported that abrupt exposure of M. edulis acclimated to 30%oto 15%osea water resulted in an increase in the rate of ‘excretion’ of amino acid (i.e. primary amines), from approximately 0 to a maximum of 3 μgNg⁻¹drymassh⁻¹. For purposes of comparison, this rate can be expressed per gram wet mass of gill tissue, or approximately 0·3 μmol amino acid g⁻¹ h⁻¹. This is probably an underestimate of the ‘initial’ rate of amino acid loss from these animals, as they were allowed to pump in the 200 ml test solution for 2h before samples were collected for analysis; our data suggest that amino acid loss in a restricted volume is not linear for 2h. Henry & Mangum (1980) reported that a hypo-osmotic shock caused a rapid, though transient, loss of amino acid from gills of the estuarine clam, Rangia cuneata. It is not clear from either of these studies whether the loss of amino acid from integumental tissues plays a quantitatively significant role in the reduction of cellular amino acid levels associated with isosmotic volume regulation in these animals.

The observation that exposure to leucine and lysine caused a specific loss of neutral amino acids (e.g. Fig. 7) is consistent with earlier observations that exposure to GABA results in the specific loss of taurine from integumental tissues (Wright & Secomb, 1986). The integumental transport of taurine and neutral amino acids involves separate transport pathways (Wright, 1985), suggesting that the efflux caused by exogenous amino acids involves interaction of the external amino acid with its specific carrier. We suggested that the increase in taurine loss to sea water upon addition of GABA was due to two separate phenomena: (i) a competitive inhibition that prevents the reaccumulation of taurine lost from the animal; and (ii) an accelerative heteroexchange of external GABA for endogenous taurine. In the present series of experiments, the rapid appearance of neutral amino acids observed upon the addition of leucine (and to a lesser extent, lysine) was probably due primarily to heteroexchange. The extremely low steady-state concentrations for the neutral amino acids are consistent with routine efflux rates that account for less than 20% of the observed alanine lost in the presence of leucine. It was of interest that alanine, though present in the gill at approximately half the concentration of glycine, was lost four times faster (e.g. Fig. 7). This could be explained if the affinity of the alanine/glycine transporter(s) at the intracellular, or trans, face of the membrane was greater for alanine than for glycine. This possibility is supported by the observation that alanine (and other lipophilic neutral amino acids) is a better inhibitor of neutral transport pathways than glycine (and other hydrophilic amino acids) (Wright, 1985).

In conclusion, the apical membrane of cells from the integumental gill epithelium has a microvillous brush border that serves to increase the apical area by approximately 13-fold, though the dimensions of these microvilli are such that they should not influence the kinetics of solute exchange across the apical membrane. The apical poles of gill cells serve as a barrier for the passive loss of endogenous amino acids to surrounding sea water. The apparent permeability of these cells to amino acids is much lower than that noted in mammalian systems. This low permeability, when combined with the presence of several separate pathways for the accumulation of amino acid from sea water, results in a low rate of loss of amino acid from the tissue, and thereby helps to maintain the very large intracellular concentrations of amino acid used by integumental cells in isosmotic volume regulation. Exposure to dilute sea water causes a general increase in the permeability of the gill to amino acids. This increase is transient and reversible, and may play a part in volume regulation.

In order to analyse the potential for reabsorption of leaked substrate within the brush border, we consider the configuration shown in Fig. 8A. Uniform cylindrical microvilli of fixed length project from a base plane. Leakage is assumed to occur in this plane only, i.e. across the apical membrane at the base of the microvilli, and reabsorption is considered to be uniformly distributed along the sides of the microvilli, and to be described by Michaelis–Menten kinetics. As already indicated, this hypothetical configuration was chosen to provide an upper boundary for reabsorption, compared to cases in which leakage is distributed more uniformly over the microvillar surface.

It is assumed that the fluid contained between the microvilli is stationary, so that solute transport occurs by diffusion and not convection. A further important simplification is made possible by the fact that the spaces between microvilli are relatively long and narrow. Concentration variations across these spaces are therefore generally negligible compared with the longitudinal variations. In this approximation, the concentration is given by C(z) (0 ⩽z ⩽z_m) where z is distance from the base plane, and z_m is the length of the microvilli. Similarly, the outward solute flux per unit area of the base plane is represented by Q(z).

For the microvilli lining the gill epithelia studied here, typical parameter values are: a_g = 0·60, J_fmax = 19·3 × 10⁻¹²molcm⁻²s⁻¹,K_t = 4·9 pmol l⁻¹, D = 10⁻³cm²s⁻¹, z_m = 1 jam. The resulting estimate is λz_m = 0·26. This value is sufficiently small that the fluid surrounding the microvilli may be considered well-mixed. However, it is clear that geometrical and transport parameters of the same order of magnitude could readily lead to values of z_m around unity. For instance, for the brush border in insect intestinal epithelium, a_g≈0·2 and z_m ≈ 4·5pm (Smith et al. 1969). If the same transport parameters are assumed as for the gill epithelia, then λz_m ≈ 0·94.

The authors express their thanks to Ms Debra Moon and Ms Margaret Coulombo for technical assistance throughout the course of this work. This study was funded by NSF grants PCM82-16745 and DCB85-17769 to SHW and PCM82-15420 to TJB.