Review of volume regulation in response to hypertonic challenge and secretagogues.

© R.J.Walters 1992. All rights reserved.

  During transepithelial solute flow epithelial cells maintain their volume in the face of a virtual torrential flood, and so must constantly readjust the transport processes mediating solute entry into and exit from the cell, and the measurement of cell volume functions as a sensitive indicator of the balance of these two processes.

  Regulation of cell volume in endothelium and epithelium is critical in the regulation of fluid and electrolyte transport between tissue compartments (e.g. intestinal lumen and villus microcirculation; cerebrospinal fluid and cranial microvasculature). In the vascular endothelium, factors that cause endothelial shrinkage (e.g. hypertonicity) or retraction (e.g. histamine) are known to increase endothelial permeability. Although large step changes in extracellular osmolarity may be uncharacteristic of the physiological and pathophysiological states that non-intestinal cells are exposed to; in the intestine it is likely that the epithelium will routinely encounter large fluctuations in luminal osmolarity and will therefore require cell mechanisms to recover changes in cell volume rapidly, as intestinal epithelial cells must coordinate not only ion fluxes that regulate volume, but also those responsible for transcellular solute transport. In other cell types studies of volume regulation have been performed more "physiologically" by exposing cells to gradual changes in extracellular osmolarity wherein they undergo isovolumetric regulation (IVR) up to a threshold osmolarity beyond which point they shrink (J.W.Lowr et al AJP 256 F622-F631 1989).

 Regulation of volume in response to hyperosmotic media.

  Cell shrinkage in response to a hyperosmotic challenge stimulates net ion and osmotically obliged water uptake in many cell types leading to a restoration of "normal" cell volume, a process known as regulatory volume increase (RVI). Cells capable of RVI in response to RVI can be distinguished into 2 categories, those that can undergo RVI in response to a direct elevation in extracellular osmolarity (e.g. necturus gallbladder cells) and those that only undergo RVI only in response to the RVD/RVI protocol; in which the cell is exposed first to a hypotonic and then to an isotonic medium (e.g. mouse pancreatic ß-cells, Ehrlich ascite tumour cells and intestinal 407 cells). Such cell types appear incapable of RVI in response to direct hypertonic challenge. In vitro in invertebrate cells nonelectrolytes play a major role in the RVI response; whereas in vertebrate cells the RVI response is accounted for predominantly by electrolytes.

  RVI is generally reported to be mediated by stimulation of either Na-K-2Cl cotransport, NaCl cotransport or Na⁺-H⁺ exchange (cite review). Na-K-2Cl cotransport systems have been defined by 3 criteria:(1) interdependency of K⁺, Na⁺ and Cl^- ions present on the same side of the membrane, (2) ion selectivity, and (3) specificity of pharmacological inhibition. An alternative mechanism that has been proposed is the simultaneous activation of NaCl and KCl symporters; however the KCl symport has been shown to be hypotonically activated (Lauf J.Memb.Biol. 88: 1-13 1985), is relatively insensitive to loop diuretics and has an anion preference of Br^-³ Cl^-, the reverse of the Na-K-2Cl cotransport system (MacLeod AJP 258 G665-G674 1990).

  We have previously demonstrated that small intestinal crypts undergo complete RVD in response to a sharp decrease in the osmolarity of the bathing medium (J.A.O'Brien et al BBA 1991); and that intact crypts recover their volume after secretagogue-induced volume decrease (SVD) upon washout of agonist (SVI). Measurements of crypt volume after exposure to hypertonic (133% original osmolarity) or hypotonic (67% original osmolarity) solutions are close to those expected for an ideal osmometer, suggesting that estimating crypt morphology to be cylindrical gives a good approximation of crypt volume. Hyperosmotic shrinkage was induced by mannitol addition to ensure that changes in cell volume did not occur as a result of alterations of the ionic composition of the bathing medium.However it has not been demonstrated whether small intestinal crypts have the capacity to recover their volume in response to hyperosmotic shrinkage (RVI) or in the maintained presence of secretagogue. In addition the mechanism of SVI, and RVI, if it occurs, have yet to be elucidated in the crypt.

  The % recovery rates per minute in the crypt in response to isosmotic (100nM VIP) and anisoosmotic (133% osmolarity) shrinkage are 4.6±0.8%/min and 4.2±0.8%/min respectively. This suggests that although the rate of SVI appears to be more rapid than RVI, notably in the first 4 minutes after agonist washout, we do have the resolution to answer this question here. The removal of all but 4.2mM extracellular Na⁺ by replacement of NaCl with N-methyl D-glucamine chloride is associated with a rapid 5% decrease in cell volume. This is probably due to the depletion of cellular Na⁺ that is known to occur through Na⁺-K⁺ ATPase activity in Na⁺-free medium when Na⁺ uptake by Na⁺-H⁺ exchange and/or Na-K-2Cl cotransport is prevented. The gradual reduction in cell volume that occurs in Cl^- free medium might be explained by a reduction in tonic Cl^- dependent Na-K-2Cl cotransport activity. Alternatively the volume reduction might result from an increased rate of Cl^- efflux from the cell through conductive pathways due to the increased electrochemical gradient resulting from extracellular Cl^- removal. However failure to observe RVI in the absence of extracellular Na⁺ or Cl^- could be interpreted as due to a lack of glucose in the medium, which would provide a metabolic substrate for transport activity (e.g. acetate dependence of Iso-Volumetric-Regulation in renal proximal tubules J.W.Lohr AJP 256 F622-F631 1989).

It is interesting that RVI occurs in small intestinal crypts in response to hyperosmotic shrinkage, but not in other cell types (e.g. pancreatic ß-cells). In pancreatic ß-cells both RVD and RVI following the replacement of hypotonic medium with isotonic medium, are effectively inhibited by 1mM furosemide (K.G.Engstrom et al BBA 1991). RVI did not occur in response to challenge with a medium made hypertonic with the addition of either 50mM NaCl or 200mM sucrose but a slow RVI did occur after prolonged exposure to hypotonicity when the cells were returned to original isotonic bathing medium. This may be due to differences in the gradient for inwardly-directed ion cotransport resulting from an increase in the final intracellular activities of ions after cell shrinkage, which in turn depends upon differences in intracellular ion activities between cell types under isotonic conditions and the magnitude and nature (i.e. elevation of extracellular NaCl or addition of membrane impermeant osmolyte such as mannitol) of the hyperosmotic "load" placed upon them.

The dependence of crypt RVI and SVI upon extracellular Cl^- ions and upon an inwardly directed Na⁺ gradient is consistent with Na-K-2Cl cotransport and NaCl cotransport, as well as with Na⁺-H⁺ exchange if a Cl^--HCO₃^- exchanger is functioning in parallel. Only a dependence on extracellular K⁺ concentration can distinguish between these 3 transport processes. It has been reported that RVI in Ehrlich Ascite tumour cells is highly dependent upon extracellular K⁺ ions; RVI being abolished at less than 2mM extracellular K⁺, at which concentration cells lose KCl and shrink. The inhibition of both RVI and SVI in the crypt by low concentrations (1m M) of the loop diuretic inhibitor bumetanide and the insensitivity of the volume recovery to either the addition of amiloride (1mM) or ethylisopropylamiloride (40m M), at concentrations known to inhibit both the sensitive and insensitive Na⁺-H⁺ exchanger subtypes completely is indicative of a Na-K-2Cl mechanism (Na⁺-H⁺ subtypes: a predictive review, J.D.Clark and L.E.Limbird, Am.J.Physiol. 261: C945-C953, 1991). However our results are in contradiction to the data of Hamilton and McLeod who reported that RVI in a dissociated preparation of guinea-pig jejunal crypt enterocytes was blocked by amiloride (?mM) but not 10m M bumetanide. A bumetanide-sensitive Na-K-2Cl cotransport activity has been reported to be present in intact crypt units isolated from rat duodenum, measured by ⁸⁶Rb⁺ uptake (C.M.McNicholas (1992) J.Physiol. 446, 6P), but the mechanism of its regulation has yet to be elucidated.

If we simplify by assuming that intracellular ion activities vary in proportion to the change in volume induced in a perfect osmometer then we can predict the initial variation in intracellular activities of individual ions when the cell is exposed to solutions of varying tonicity. If the ß-cell is exposed a solution of 163% original tonicity (by the addition of nonelectrolyte), then we might expect the intracellular ion activities to increase proportionately to the shrinkage in volume. The increased intracellular Na⁺, K⁺ and Cl^- concentrations that result might effectively reverse the direction of net ion co-transport which is determined largely by the transmembrane gradients of these transported ions. A similar exposure to medium of 163% original osmolarity may similarly inhibit RVI in the small intestinal crypt, in which case the gradient for net cotransport still appears to be inward following exposure to a medium 133% of original osmolarity. This argument may explain why RVI occurs in ß-cells pre-equilibrated in hypotonic medium upon exposure to medium of original osmolarity but not upon direct exposure to hypertonic media. If we assume initial intracellular ion activities in the ß-cell to be 100; 60 and 10 mM for K⁺; Cl^- and Na⁺ respectively with a relative volume (RV) of 1.0 then exposure to medium of 167% relative osmolarity (RO) will result in an RV of 0.6 and initial intracellular ionic activities of 167; 100 and 17 respectively. However if the cell is exposed first to medium of 83% RO, resulting in a RV of 1.2 and initial intracellular ion activities of 83; 50 and 8 and allowed to undergo RVD until the RV returns to near 1.0; then assuming that K⁺ and Cl^- ions are lost in equimolar proportions to restore initial volume we may assume that only small changes in intracellular ion activities subsequently occur. If "isotonic" (relatively hypertonic) medium is restored, a rapid initial return to the original RV of 0.83 will be accompanied by the restoration of intracellular ion activities of 100; 60 and 10mM respectively; considerably less than those predicted to result from exposure to 167% hypertonic medium, thus maintaining a gradient favourable to inwardly-directed cotransport. We might conclude that cells that undergo RVI in response to the hypotonic-isotonic protocol but not on exposure to hyperosmotic media may still effect recruitment of additional cotransporters but net ion uptake may be prevented by a reversal of the transmembrane ionic gradient for uptake. In duck erythrocytes if hyperosmotic conditions are established by the addition of impermeant molecules, rather than Na/KCl, without supplementation of extracellular K⁺ above 5mM, then RVI does not occur, as the sum of the chemical potential gradients for Na⁺, K⁺ and Cl^- are close to zero at low [K_o]. However this does not appear to be the case for intestinal crypts wherein the addition of mannitol is adequate for complete RVI to be induced, possibly reflecting a comparatively low resting intracellular concentration of Cl^- and/or K⁺ ions in crypt enterocytes.

We have demonstrated that crypt RVD and SVD are driven by the parallel efflux of

K⁺ and Cl^- ions from the cytosol in response to cell swelling and increases in the adenylate cyclase or phospholipase C mediated signalling pathways respectively. Whether the increase in ionic conductance in RVD is due to membrane distention, Ca²⁺ influx or another signalling pathway (e.g. lipoxygenase metabolic products) has yet to be elucidated. As we have demonstrated that the crypt resting membrane potential is dominated by a basolateral K⁺ conductance it may be reasonable to speculate that it is the activation of anionic permeability pathways that is rate limiting for KCl efflux in RVD and SVD, however this remains to be demonstrated directly. If we assume in the resting crypt that steady state volume results from an equilibrium between ion uptake and ion efflux pathways, then for SVD and RVD to occur then the rate of ion loss through permeability pathways must exceed the rate of ion uptake. However we have demonstrated previously that the crypt membrane conductance is elevated for the duration of secretagogue application. Therefore if, after 8 minutes, a new steady state is maintained at a reduced cellular volume after isoosmotic shrinkage, we can therefore argue that the rate of ion uptake must also be increased for this second steady state to occur. This must require either an increase in the number of ions taken up per transporter or an increase in the number of transporters active in the membrane. If K⁺ or Cl^- permeabilities are reduced upon secretagogue washout, then the volume loss is recovered as the rate of ion uptake now exceeds the rate of ion loss. In eccrine clear cells methacholine induced shrinkage is reversed upon agonist washout, but in the continued presence of high concentration of agonist (3m M) most cells (77%) exhibited a gradual volume recovery, whilst 23% showed a maintained shrinkage for the duration of agonist addition (Suzuki et al J.Memb.Biol. 123. 33-41 1991). The SVD, which was as much as 30% of resting volume in some cells is of similar magnitude to the mean shrinkages induced by VIP (25%) and carbachol (33%) in the crypts.

It is proposed that the continued progressive shrinkage observed in pancreatic ß-cells in response to the addition of 200mM sucrose is due to a net ion efflux via a furosemide-sensitive Cl^--cation cotransport mechanism (K.G.Engstrom et al BBA 1991). In chinese hampster ovary cells (CHO cells) hyperosmotic shrinkage activated bumetanide sensitive cotransport, but this did not contribute to RVI, as this resulted in the stimulation of Na-K-2Cl efflux (Rotin et al AJP 257 C1158-C1165 1989).

Bidirectionality is a well-characterised property of Na-K-2Cl cotransport (O'Grady AJP 253 C177-C192 1987), consequently at low extracellular K⁺ (<0.2mM) the system operates in efflux and cells shrink. This might explain the progressive shrinkage in hypertonicity seen in the crypt when bathed in low extracellular Na⁺ or Cl^-. In fact observations from a wide variety of non-intestinal cell-types indicate that intracellular Cl^- must be below a threshold level before shrinkage can activate RVI, a situation created by the RVI after RVD protocol; although this is not necessarily the case for intestinal crypt or villus enterocytes which can recover from a direct anisoosmotic load produced by a solution 154% of original osmolarity.

The cell membrane potential was monitored in experiments in which the osmolarity of the bathing medium was increased from 300 to 400 milliosmoles by the addition of 100mM mannitol. An initial depolarisation (mean 10±2mV) of 1-3 min duration was associated with a small apparent increase in membrane conductance (» 10%). Within 10-20 minutes of increasing bath osmolarity E_m reached a new steady state 7.6 ±1mV more hyperpolarised than initial E_m. This hyperpolarisation was slowly oscillating and gradual and was accompanied by a large progressive increase in membrane conductance. The hyperpolarisation phase but not the depolarisation phase of the response to hypertonicity was

reversibly inhibited by 50m M bumetanide.

The loss of cytosolic water resulting from an increase in bath osmolarity would increase the intracellular activities of Na⁺, K⁺ and Cl^- ions as well as increasing the concentration of non-permeable organic osmolytes. As intracellular [Cl^-] increases then E_Cl would become more positive, depolarising E_m. Alternatively the depolarisation could be interpreted by an influx of Na⁺ ions or by the inhibition of a membrane-tension dependent K⁺ conductance, but the latter would not explain the slight increase in membrane conductance consistently seen. RVI in aortic endothelial cells is associated with a large initial increase in cell Na⁺ activity within the first 5 minutes which then recovers towards the pre-shrinkage value (O'Neill et al AJP 1992). Depolarisations have been reported in response to hypertonic stress in intestinal 407 cells (Okada NIPS Vol4 Dec 1989 238-242) and gallbladder epithelial cells (L.Reuss PNAS 82: 6014-6018; 1985). Similar depolarisations have been reported in response to the hypo-iso protocol in Int.407 cells and in MDCK cells (Roy et al J.Memb.Biol. 100: 83-96 1987), which in the Int.407 cells has been attributed to a Na⁺ conductance. Such a depolarising entry of Na⁺ ions would favour Cl^- ion entry via Cl^- channels activated by prior hypotonic shock, and this is proposed to be the mechanism of RVI in MDCK cells under these conditions.

The hyperpolarisation phase, but not the large conductance increase, could be explained by a reduction in Cl^- permeability. However the gradual hyperpolarisation of membrane potential could be reasonably explained by a progressive accumulation of K⁺ ions intracellularly due to the parallel operation of the bumetanide-sensitive co-transport and Na⁺-K⁺ ATPase activities, the latter possibly activated by an increase in intracellular Na⁺ activity. The large increase in conductance could be explained by an increase in K⁺ recycling through the constitutively active basolateral K⁺ conductance as the electrochemical gradient for K⁺ exit would be expected to increase with its intracellular activity. This would also account for the movement towards E_K. In rat pancreatic ß-cells exposure to medium made hypertonic by the addition of 200mM sucrose is associated with a 40% increase in ⁸⁶Rb⁺ efflux, and this increase is abolished by the addition of 1mM furosemide (K.G.Engstrom et al BBA 1991). Similar observations have been made in cultured HeLa cells where addition of mannitol to the bathing medium resulted in an increase in loop-diuretic insensitive ⁸⁶Rb⁺ efflux persistent with the hypertonic stimulus.

The mechanisms by which cells sense volume loss and transduce these signals to the modulation of ion transport remains to be elucidated. Two hypotheses of how volume sensitive transport mechanisms are regulated predominate; (i) a mechanical strain sensor in the cell membrane or its associated cytoskeleton modulates the transporter and (ii) that cell volume changes alter the intracellular concentration of a soluble moiety initiating events that lead to transporter activation. Okadaic acid activates Na-K-2Cl cotransport in human red cells even in the absence of cell shrinkage (Pewitt et al J.Biol.Chem. 265, 20747-20756 1991). The principle is that okadaic acid by blocking phosphatase activity leads to the progressive phosphorylation of some regulatory component by kinase activity, and furthermore protein kinase inhibitors inactivate cAMP induced cotransport. Thus stimulation by cell shrinkage increases cell phosphorylation relative to dephosphorylation so that fractional phosphorylation increases and the cotransporter is progressively activated. In osmotically shrunken lymphocytes increased protein and phospholipid phosphorylation occurred, with an associated increase in turnover of phosphoinositides that was not as a result of PKC activation, and might possibly be due to the activation of a phosphoinositide kinase (S.Grinstein J.Memb.Biol. 90: 1-12 1986).

A greater stimulation of Na-K-2Cl cotransport was observed in aortic endothelial cells when shrunken isoosmotically than when shrunken hyperosmotically (W.C.O'Neill et al, Am.J.Physiol. 262 C436-C444, 1992). An increase in bumetanide-sensitive K⁺ influx was observed upon hypertonic shrinkage in the aortic endothelial cell. Increased bumetanide-sensitive K⁺ influx was associated with an increased number of [³H]bumetanide binding sites rather than an increased influx per binding site in shrunken cells.

O'Neill et al (AJP 262 C436-C444, 1992) have proposed that changes in cell volume alter the number of functional cotransporters in the membrane either by the recruitment of preformed cotransporters to the membrane or due to the activation of latent transporters incapable of either transport or bumetanide binding. They also suggest that cotransport activity is regulated by trans-inhibition by intracellular Cl^-, first proposed to occur in the squid axon (Breitweiser et al AJP, 258, C749-C753, 1990). Interestingly Na-K-2Cl cotransport is also activated by raising intracellular cAMP in avian red blood cells in the absence of cell shrinkage (Kregenow Annu.Rev.Physiol. 43: 493-505 1981) and by vasoactive peptides in endothelial cells (O'Donnell AJP 257 C39-C46 1989). However Na-K-2Cl cotransport was stimulated in Chinese Hampster Ovary cells but did not mediate a net influx of ions. RVI in these cells was mediated instead by Na⁺-H⁺ exchange (Rotin et al AJP 257 C1158-C1165 1989). MacLeod and Hamilton concluded that in villus enterocytes under isotonic, isoosmotic conditions that the Na-K-2Cl cotransporter is active, but that at equilibrium does not contribute to the maintenance of steady-state volume and thus bumetanide has no effect on cell volume.

Cell shrinkage has been reported to stimulate Na⁺-H⁺ exchange via the PKC-independent activation of a G-protein in barnacle muscle fibres (B.A.Davis et al, Am.J.Physiol.1992). It will be interesting to determine if G-proteins are involved in volume shrinkage signal transduction in other cell types, including those in which RVI is mediated by transport systems other than Na⁺-H⁺ exchange.

In small intestinal crypts the mechanism of activation of ion uptake in RVI and SVI in crypts in response to hypertonic or secretagogue challenge appear to be the same. Secretagogues induce volume decrease under iso-osmotic conditions, in which intracellular ion activities may change appreciably, particularly if a non-selective cation conductance is activated. Hypertonic (anisoosmotic) challenge will result in a large associated increase in intracellular ion activities, although its effects upon phospholipase C and adenylate cyclase mediated signalling pathways are unknown in the crypt. If SVI induced upon carbachol washout is also mediated by Na-K-2Cl cotransport it would seem less likely that increases in intracellular Ca²⁺ or cAMP are the specific signalling pathway involved in the mechanism of signal transduction.

There appears to be no apparent correlation between the physiological function of a cell type and the transport mechanism (i.e. Na-K-2Cl cotransport or Na⁺-H⁺ exchange) employed in RVI. Of all cell types capable of RVI that have been characterised above, those that perform a tissue compartment interface/ barrier function (i.e. endothelial, epithelial and renal cells) show no distinct tendency to fall into either category. However fewer cell types mediating RVI principally by a Na-K-2Cl cotransport system (3/8) are able to do so without prior exposure to a hypotonic medium (the RVD before RVI protocol) than those employing a Na⁺-H⁺ exchange mechanism (6/8), possibly reflecting a greater capacity of the Na⁺-H⁺ and Cl^--HCO₃^- antiports to accumulate Na⁺ and Cl^- ions against a transmembrane gradient, driven by the carbonic anhydrase and the Na⁺-K⁺ ATPase activities. A study of the rates of RVI in various cell types again shows no clear difference between cell types or transport mechanism employed. There is however a considerable variation in RVI rates from the slowest (mouse pancreatic ß-cell 0.33% change in volume per min) to the fastest (intestinal 407 cell line 9.5%/min). In most cell types tested the rate of RVI is commonly between a 1 and 2% change in cell volume per minute.

The molecular mechanisms coupling volume loss to the activation of ion transport in epithelial and non-epithelial cell types is essential to our understanding of cell volume homeostasis. Future strategies may involve the cloning of the epithelial Na-K-2Cl cotransporter and its introduction into cell types known not to express it; or the introduction of mRNA into the Xenopus oocyte expression system and determining the effect of various stimuli upon cotransporter mediated ion fluxes.

If the Na-K-2Cl cotransporter is indeed activated in response to cell shrinkage, it might be reasonable to assume that this could be detected as a increase in the bumetanide sensitive component of ⁸⁶Rb⁺ uptake into crypt units in response to exposure to secretagogue or hypertonicity. A large increase in the initial rate of bumetanide sensitive ⁸⁶Rb⁺ influx was induced by hypertonic shrinkage in guinea pig jejunal villus cells (R.J.MacLeod and J.R.Hamilton AJP 258, G665-G674, 1990.)

If the secretagogue induced shrinkage of isolated crypt diameter prevents a visible expansion of lumenal volume, then in the intact epithelium where the crypt is firmly anchored to the basement membrane SVD might well be associated with a discernible expansion of the lumenal space, which might represent an adaptive mechanism for the increased fluid flow evoked in intestinal secretion.