AQP and the Control of Fluid Transport in a Salivary Gland

Abstract

Experiments were performed with the perfused rat submandibular gland in vitro to investigate the nature of the coupling between transported salt and water by varying the osmolarity of the source bath and observing the changes in secretory volume flow. Glands were submitted to hypertonic step changes by changing the saline perfusate to one containing different levels of sucrose. The flow rate responded by falling to a lower value, establishing a new steady-state flow. The rate changes did not correspond to those expected from a system in which fluid production is due to simple osmotic equilibration, but were much larger. The changes were fitted to a model in which fluid production is largely paracellular, the rate of which is controlled by an osmosensor system in the basal membrane. The same experiments were done with glands from rats that had been bred to have very low levels of AQP5 (the principal aquaporin of the salivary acinar cell) in which little AQP5 is expressed at the basal membrane. In these rats, salivary secretion rates after hypertonic challenges were small and best modelled by simple osmotic equilibration. In rats which had intermediate AQP5 levels the changes in flow rate were similar to those of normal rats although their AQP5 levels were reduced.

Finally, perfused normal glands were subject to retrograde ductal injection of salines containing different levels of Hg²⁺ ions (0, 10 and 100 μM) which would act as inhibitors of AQP5 at the apical acinar membrane. The overall flow rates were progressively diminished with rising Hg²⁺ concentration, but after hypertonic challenge the changes in flow rates were unchanged and similar to those of normal rats.

All these results are difficult to explain by a cellular osmotic model but can be explained by a model in which paracellular flow is controlled by an osmosensor (presumably AQP5) present on the basal membrane.

Appendix

THE OSMOSENSOR-JFT MODEL AND THE SALIVARY EPITHELIUM

The model presented here is one that has been suggested previously [26] and is developed for a forward-facing epithelium in this issue [12] in which the behavior has been explored. Here we apply the principles with a minimum of adaptation to the SMG gland, a backwards-facing epithelium that transports from serosa to lumen (Fig.  A1). Two forms of the model have been developed, a steady-state solution and a time-dependent system used to investigate the changes that occur when a perfused gland in a steady state is subjected to a step change. Central to the model is the JFT (junctional fluid transfer) system which is considered to be the basis of fluid production in isotonically transporting epithelia [26]. This fluid transfer pathway is located in the junction but is functionally different from the leak pathway with which it may be considered to be in parallel. The JFT system implies the coupled transfer of water and solute, in particular the NaCl of the bathing saline. The pathway has a selectivity θ which represents the ratio of salt:water transferred compared to the ratio in the source bath. The JFT rate is controlled by a membrane osmosensor, ideally located at the membrane separating source bath and cell, here considered to be AQP5 at the basal membrane.

Figure A1

Major features of the sensor-feedback model applied to SMG. Feedback control of the JFT system is proposed for the basal AQP5 acting as an osmosensor.

STEADY STATE

The feedback model is described by three linear simultaneous equations

$$ {jv_{p} = Am{\left( {C_{\rm c} - C_{\rm b} } \right)} + jvo } $$

(A1)

$$ {C_{\rm c} = {\left( {P_{\rm a} A_{\rm a} C_{\rm a} + P_{\rm b} A_{\rm b} C_{\rm b} } \right)}/{\left( {P_{\rm a} A_{\rm a} + P_{\rm b} A_{\rm b} } \right)}} $$

(A2)

$$ \eqalign{C_{\rm c} &= {\left( {jv_{\rm p} C_{\rm b} {\rm{\theta }} + P_{\rm s} {\left( {C_{\rm b} - C_{\rm a} } \right)} + js_{\rm c} } \right)}/{\left( {jv_{\rm p} + P_{\rm a} A_{\rm a} {\left( {C_{\rm a} - C_{\rm c} } \right)}} \right)}\cr &= J_{\rm s} /J_{\rm v}.} $$

(A3)

The basic parameters in these equations are:

• jv _p junctional volume flow cm s⁻¹

• jvo junctional volume flow offset cm s⁻¹

• js _c trans-cellular pumping rate osmole cm⁻²s⁻¹

• A gain of the feedback loop cm⁴ osmole⁻¹s⁻¹

• m sensor density in the membrane

• C _a,b,c apical, basal or cell osmolarity osmole cm⁻³

• P _a,b apical, basal osmotic permeabilities cm⁴osmole⁻¹sec⁻¹

• A _a,b apical, basal membrane areas cm²(cm²epithelium)⁻¹

• P _s solute permeability of the leak pathway cm s⁻¹

• θ selectivity of the JFT pathway

Eq. A1 describes the transduction of the osmotic signal input by a sensor and signaling sequence at the basal membrane into a junctional volume flow jv _p in addition to the offset, i.e., the rate at which the JFT transfers volume in the absence of the signal. In this paper jvo is set to zero. The sensor output depends upon its density m in the membrane and has a gain A.

Eq. A2 describes the osmotic concentration in the cell, which is the result of osmotic flow over the two membranes being equal in the steady state to preserve cell volume.

Eq. A3 describes the emergent fluid osmolarity, given as the ratio of transepithelial salt flow J _s to the overall fluid transport rate J _v . Js is made up of three components. (i) jv _p C _b θ the convection of salt through the JFT driven by the fluid transport mechanism with a selectivity θ. (ii) P _s (C _b − C _a ) the salt diffusion between baths across the paracellular system where P _s is the overall permeability. In this paper P _s is set to zero because the leak pathway, although permeable to individual ions, has a low overall permeability to salt. (iii) js _c the rate of active salt pumping through the trans-cellular route. J _v is made up of two components. (i) the paracellular volume flow jv _p and (ii) the osmotic flow from cell to luminal space, P _a A _a (C _a − C _c ).

The parameters were given a standard set of initial values approximating to a general epithelial cell. In many transporting epithelia the apical membrane is much smaller than the basolateral membrane and this certainly applies to the acinar cells of most exocrine glands. Although the acinar epithelium is topologically spherical or tubular, the model has no geometry but the cell and interspace are given area and volume. This is not an important criterion because the dominant feature of the cell turns out to be the overall osmotic conductance of each membrane P _a A _a and P _b A _b and especially their ratio. The values for A _a and A _b used here are 1 and 10. The osmotic permeability is 2 × 10⁻² for both membranes, close to the value for red cells. Salt is pumped across the cell by the general mechanism for backward-facing epithelia in which Na recirculation at the basal membrane drives the extrusion of Cl through the apical membrane [16]. The rate js _c for this was 2 × 10⁻⁹ mole s⁻¹ per cm⁻² epithelium, a moderate value for secretory epithelia. In this feedback model the general behavior is not affected by the magnitude of js _c and under isotonic conditions the pumping merely determines the overall fluid transport rate.

One of the most important parameters is θ the selectivity. Changes in this affect the proportion of salt passing the JFT but not the tonicity of the fluid. The selectivity is also equal to the fraction of salt crossing the epithelium by the paracellular route under normal conditions. The paracellular salt flow is jv _p C _b θ and the transepithelial salt flow is equal to J _v C _b , so this fraction is given by jv _p C _b θ/J _v C _b . As the system is functioning close to isotonicity then virtually all the volume flow is paracellular so that jv _p = Jv and the paracellular salt fraction becomes equal to θ.

Finally, the gain A is the overriding parameter (strictly Am, as m = 1 here). We have no precise knowledge of this until we understand the coupling between the sensor and the JFT. However, when A is increased above zero we very quickly enter a state of ‘osmotic clamp’ where the relative osmolarity Os becomes close to 1.0 and the transported fluid may be regarded as isotonic. In this condition the osmotic permeabilities of the membrane play little part and most of the fluid flow is paracellular. This isotonic condition is very stable and is independent of changes in the permeabilities, the geometry and the salt pumping rate of the acinar cells.

TIME-DEPENDENT VERSION

This version has been used to follow changes after hypertonic challenge shown in Fig. 3 by incorporating time-dependent elements. Briefly, the rise of concentration change adjacent to the basal membrane after switching salines is treated as a sub-epithelial diffusive compartment with a single exponential time constant λ with a half-time of 20 s. The osmosensor system has also an exponential time constant k with a half-time of 10 s. This represents the reaction of the sensing molecule (probably AQP5) and the cell signalling chain to the JFT system. These are estimates that can be justified on simple physical grounds. Finally, there is a volume V _lis assigned to the lateral interspace system around the cell, 0.5 μm wide and of cell height 25 μm. Water flows are calculated over basal membrane, interspace membrane and apical membrane. The concentration of salt in the lateral interspace varies during transients but is dominated by C _b in the steady state.

As the cells shrink during hypertonic challenges the volume change will be distributed between lumen and the sub-epithelial side of the cell. The luminal volume shift will decrease the apparent secretion rate but this effect will decline as the cell equilibrates. The interspace concentrations are calculated in discrete time steps which determine the rate of salt transport by the JFT system and the osmotic exchanges with the cell. The cell volume at every step of the evolution is calculated and partitioned symmetrically using a factor β = 0.5 between basal and apical sides of the cell. This procedure is straightforward and reproduces the swings of secretion rate between the steady states quite well. When β is set to zero the undershoot of Fig. 3 is not seen.

The time step lengths were systematically explored and the curve of Fig. 3, which is similar in magnitude and time to the experimental data of Fig. 2, was generated by using the above values of β, λ and k. Damped oscillations on entry to the new steady state could be generated by adjusting the values of λ and k as might be expected for a feedback system. These oscillations can be seen in many experiments during osmotic challenge. Together with other transients, they are indications that a feedback system is in operation and remain to be explored further.