V.POTENTIAL UNIDIRECTIONAL REABSORPTION OF AQUEOUS HUMOR

A.Transport Components Underlying Potential Transcellular Reabsorption Across the Ciliary Epithelium

In addition to mechanisms supporting transcellular transfer of solute and water (Fig. 1A), a number of transporters have been identified that can underlie translocation of fluid in the opposite direction (Fig. 1B). At the aqueous surface, NaCl may be reabsorbed by Naþ/Hþ and Cl /HCO3 antiports, Naþ Kþ 2Cl and Naþ Cl symports, and amiloride sensitive Naþ channels (Crook et al., 1992; Von Brauchitsch and Crook, 1993; Crook and Polansky, 1994; Dong and Delamere, 1994; Civan et al., 1996; Crook and Riese, 1996; Riese et al., 1998) functionally identified in cultured NPE cells. The AQP1 and AQP4 channels at the basolateral membranes of the NPE cells (Hamann et al., 1998; Yamaguchi et al., 2006) can subserve water movement back into the cells from the aqueous humor. The fluid reabsorbed can be transferred back to the PE cells through the gap junctions linking the two cell layers.

Once the reabsorbed fluid reaches the PE cells, mechanisms are also in place for subsequent solute release into the stroma. Albeit less numerous in the PE cells (Usukura et al., 1988), Naþ, Kþ activated ATPase is expressed at the stromal, as well as at the aqueous, surface (Krupin et al., 1984; Ghosh et al., 1990; Ghosh et al., 1991). Thus, Naþ can be extruded by the PE cells back into the stroma, in parallel with Cl channels. At least one population of these PE cell channels comprises maxi Cl channels that can be synergistically activated by ATP and tamoxifen (Mitchell et al., 2000). The eVect of ATP appears mediated by stimulating cAMP (Fleischhauer et al., 2001) that acts directly on the channels (Do et al., 2004a). As illustrated by Figs. 2 and 3, the cAMP increases open channel probability at physiological membrane potentials. This eVect is larger when the PE cells have higher concentrations of intracellular Cl , which would enhance their ability to cope with increased rates of reabsorptive Cl delivery from the NPE cells. The maxi Cl channels are also activated by swelling (Zhang and Jacob, 1997), which might result from delivery of reabsorbed aqueous humor transferred via the NPE cells.

As discussed in greater depth in Chapter 4 (Macknight and Civan, 2008), electron microprobe analysis suggests that the relative importance of the potential reabsorptive pathway varies across diVerent regions of the rabbit ciliary epithelium. The physiological importance of regional transcellular reabsorption has not yet been defined. However, Naþ reabsorbed at the aqueous surface is now known to be a major determinant of the PE cell Naþ content in the anterior region of the intact rabbit ciliary epithelium (McLaughlin et al., 2007).

cAMP, 130 mM Cl−

cAMP, 65 mM Cl−

FIGURE 3 Vm dependence of open probability (Po) for maxi Cl channels in the presence of 500 mM cAMP (Do et al., 2004a). Averages were calculated from patches that displayed open events at all applied voltages. The channel displayed Vm dependent inactivation, especially when Vm was either greater than þ40 mV or smaller than 40 mV. The topmost curve represented the baseline conditions in which Cl concentrations in the micropipette and bath were 130 mM. Reducing the cytoplasmic Cl concentration from 130 mm to either 65 or 30 mm reduced Po at all potentials. The extracellular NaCl concentration was constant at 130 mM, whereas the cytoplasmic Cl concentration was varied. Curves were fitted to two Boltzmann equations. Reprinted with the permission of the American Physiological Society.

recently been reported to accelerate inflow into the mouse eye (Bakall et al., 2008). Best2 is associated with Cl currents, but its potential physiological role is unclear, in part because it also appears to facilitate outflow of aqueous humor from the eye (Zhang et al., 2008). In the absence of a comprehensive hypothesis, four regulatory pathways, which have received particular attention, are considered here.

A. Swelling Activation of Cl Channels

Over periods of minutes, swelling activation of Cl channels may be the dominant mechanism for ensuring that release of NaCl and water into the aqueous humor by the NPE cells match stromal fluid delivery through the PE cells. For example, whole cell Cl currents of isolated NPE cells can be

increased 40 fold by swelling activation (e.g., Do et al., 2005). The orientation of these channels in the intact epithelium does support transepithelial secretion of Cl . Bathing both surfaces of the isolated bovine ciliary epithelium with hypotonic solution triggers a large increase in short circuit current (Fig. 4) that can be inhibited by Cl channel blockers or by leaching Cl

Time (min)

Time (min)

FIGURE 4 EVects of bilateral hypotonicity on electrical parameters in native bovine ciliary epithelium (Do et al., 2006). (A) Measurement of transepithelial PD. Constant current pulses (3 s) of 10 A were applied to the preparation every 5 min, and the deflections (~PD) were recorded as an index of R. Isc was calculated from the measured PD and R. The aqueous surface was negative to the stromal surface. (B) The calculated Isc from the preparation of (A). Reprinted with the permission of the Association for Research in Vision and Ophthalmology.

from the tissue (Do et al., 2006). The time course of the swelling activated transepithelial current is closely similar to that of the regulatory volume decrease of the isolated bovine NPE cells (Fig. 5). Cl channel activity is enhanced by cell swelling, and thereafter returns to baseline values once release of Kþ, Cl , and water restores the initial cell volume. The importance of aqueous surface Cl channels is supported by reports that blocking their activity with NPPB inhibits both net Cl transfer by the isolated bovine ciliary epithelium (Do and To, 2000) and aqueous humor inflow by the isolated, arterially perfused bovine eye (Shahidullah et al., 2003). Increased transfer of fluid from the PE cells is expected to swell the NPE cells transiently, thereby activating the Cl channels at the aqueous surface (Section IV.B.3.b) and stimulating secretion across much of the ciliary epithelium. In those regions of the ciliary epithelium that may possibly reabsorb aqueous humor, delivery of fluid from the NPE cells is expected to trigger swelling activation of PE cell Cl channels (Section V.B), thereby reducing net secretion.

FIGURE 5 Responses of total calcein fluorescence to anisosmotic changes in volume (Do et al., 2006). Fluorescence, normalized to the baseline value in isotonic solution, increases with cell swelling. The regulatory volume decrease (RVD) after hypotonic challenge was markedly inhibited by the Cl -channel blocker NPPB. Reprinted with the permission of the Association for Research in Vision and Ophthalmology.

B. Cyclic Adenosine Monophosphate

Antagonists of b adrenergic receptors lower IOP, and topical nonselective b adrenergic antagonists have long been mainstays of glaucoma therapy (Toris and Camras, 2008). Agonists to all three b receptors, b1, b2, and b3, stimulate adenylyl cyclase to produce cAMP, an eVect mediated by the heterotrimeric G protein Gs (HoVman et al., 1996). The b blockers both reduce cAMP production and lower IOP by reducing inflow of aqueous humor. A causal relationship between these two actions has been widely presumed. However, many observations, summarized elsewhere (Yorio, 1985; McLaughlin et al., 2001a; Do and Civan, 2004), have seemed at odds with the idea that the inflow reduction by b antagonists is necessarily mediated by a fall in intracellular concentration of cAMP. Particularly puzzling have been the reports that increasing cAMP by directly stimulating adenylyl cyclase with forskolin actually lowers inflow (Caprioli et al., 1984; Lee et al., 1984), and that the b agonist isoproterenol, also expected to increase cAMP, lowers IOP in water loaded rabbits (Vareilles et al., 1977). In addition, as discussed in Chapter 4 (Macknight and Civan, 2008), application of cAMP also does not reverse the eVects of the b blocker timolol on the intracellular elemental composition of intact rabbit ciliary epithelium (McLaughlin et al., 2001a).

In part, the unexpected observations concerning the eVects of cAMP and b adrenergic agents may reflect the multiple actions of the second messenger on sites within the ciliary epithelium (Do and Civan, 2004; Table II). Several known eVects of cAMP are indeed expected to stimulate aqueous humor formation, including (Fig. 1A) activation of the Naþ Kþ 2Cl PE cell symports (Crook et al., 2000) and of some the NPE cell Cl channels (Chen et al., 1994, Edelman et al., 1995). In addition, the b adrenergic agonist isoproterenol has been observed to increase Naþ, Kþ activated ATPase activity in cultured human NPE cells (Liu et al., 2001). In contrast, direct application of cAMP can reduce net secretion (Fig. 1B) by inhibiting the Naþ, Kþ pump (Delamere and King, 1992), by blocking PE–NPE gap junctions (Do et al., 2008), and by activating maxi Cl channels of the PE cells (Fleischhauer et al., 2001; Do et al., 2004b). Given these opposing actions of cAMP on ciliary epithelial secretion, the consistently ocular hypotensive eVect of b blockers raises the possibility of compartmentation of cAMP. This possibility has been substantiated in Calu 3 cells. Huang et al. (2001) found that 1 mM adenosine increased local cAMP concentration enough to activate CFTR Cl channels with little increase in the total cAMP content. Taken together with additional results, these authors concluded that clustering of receptors, G proteins, adenylyl cyclase, and PKA permitted local activation of the target, CFTR.

The immediately foregoing considerations suggest that part of the apparent inconsistencies in the results obtained with b agonists, b antagonists, and cAMP may reflect drug triggered eVects on cAMP production in the local

PE, pigmented ciliary epithelial; NPE, nonpigmented ciliary epithelial.

microenvironment of the adrenergic receptors. In addition, cAMP does not mediate all of the actions of b adrenergic agonists (Torphy, 1994). A number of reports have recently documented that b adrenergic receptors can couple to Gi proteins, and not exclusively to Gs proteins (Denson et al., 2005). For example, Denson et al. (2005) found that the b agonist isoproterenol activates BK potassium channels by coupling to Gi, activating cytosolic phospholipase A2 (c PLA2), and stimulating production of arachidonic acid. Isoproterenol’s action was blocked by the b antagonist propranolol. It is entirely possible that the isoproterenol triggered activation of NPE cell BK channels is also mediated by arachidonic acid. Stimulation of BK channels of rabbit native NPE cells by isoproterenol is not mediated by cAMP, but does depend on G protein coupling (Bhattacharyya et al., 2002). Furthermore, arachidonic acid has long been known to activate NPE cell Kþ channels (Civan et al., 1994).

In summary, b blockers eVectively lower ciliary epithelial secretion, IOP, and cAMP formation. However, discordant results obtained by applying b agonists, b antagonists, and cAMP have raised the possibility that changes in total cellular cAMP concentration do not necessarily mediate the drug triggered changes in aqueous humor dynamics. Recent studies have now led to at least two possible explanations. First, cAMP exerts many, sometimes opposing, eVects on ciliary epithelial secretion (Table II). Administration of large concentrations of membrane permeant forms of cAMP is likely to aVect all of these transport targets. In contrast, drugs, hormones, and biologically active peptides that bind to receptors at specific membrane areas may elevate cAMP in circumscribed microenvironments, targeting a narrow

range of membrane transporters. Second, although b agonists have been widely presumed to act solely through Gs mediated production of cAMP, at least one alternative pathway has been demonstrated. The agonists and antagonists can also trigger Gi mediated activation of phospholipase A2, enhancing arachidonic acid formation.

C. Carbonic Anhydrase

Inhibition of CA provided the first successful approach for lowering IOP by reducing the rate of aqueous humor inflow (reviewed by Brubaker, 1998). The first successful clinical trials were reported more than half a century ago and the inhibitor acetazolamide has been long known to reduce accessibility of plasma HCO3 to the aqueous humor (Maren, 1976). Nevertheless, understanding of the probable mechanism of action of CA inhibitors has developed much more recently (Helbig et al., 1989a; Wiederholt et al., 1991). As discussed in Secton III. B.1.b, CA directly stimulates (Sterling et al., 2001; Li et al., 2002) the NHE1 Naþ/Hþ and AE2 Cl /HCO3 antiports (Fig. 1A; Counillon et al., 2000). Thus, CA inhibitors, such as acetazolamide and dorzolamide, likely block the first step in aqueous humor formation by inhibiting NaCl uptake from the stroma.

This hypothesis has been supported by measurements of IOP in living mice during topical inhibition of the symports (Avila et al., 2002a). Measurements were conducted with an electrophysiological approach (the servo null micropipette system) that permits continuous monitoring of IOP in the small mouse eye (Avila et al., 2001a). Topical application of each of three selective inhibitors of Naþ/Hþ antiports (Figs. 6A, and 7A, B, and D) reduced IOP. The promptness of the IOP response likely reflects enhanced delivery of drug from the tear film into the aqueous humor around the tip of the exploring micropipette (Wang et al., 2007). Bumetanide alone had no significant eVect during the period of recording (Fig. 6B). However, bumetanide further lowered IOP if applied after either the selective Naþ/Hþ exchange inhibitors (Figs. 7A, B, and D) or after blocking CA with dorzolamide (Fig. 7C). These data are consistent with the notion that the Naþ/Hþ and Cl /HCO3 antiports play a major role in secretion, and that CA inhibitors act on these exchangers to slow aqueous humor formation.

D. A3 Adenosine Receptors

Among other potential regulators of aqueous humor dynamics, A3 subtype adenosine receptors (A3ARs) are of particular interest since knockout of these receptors reduces the IOP of living mice (Avila et al., 2002b). Furthermore,

for A3ARs is expressed by cultured human NPE cells and the ciliary processes of rabbit (Mitchell et al., 1999). The similarity of the macroscopic current characteristics of the A3AR and swelling activated Cl currents suggests that both currents permeate the same channels (Carre´ et al., 2000).

Adenosine can be physiologically delivered to the aqueous surface by ATP release and ectoenzymatic metabolism of ATP by the NPE cells themselves (Mitchell et al., 1998; Fig. 8). ATP can also be released to the stromal surface by the PE cells. Binding of ATP to P2Y2 receptors (Shahidullah and Wilson, 1997) initiates a cascade leading to direct stimulation of maxi Cl channels (Fleischhauer et al., 2001; Do et al., 2004b). Tamoxifen synergistically enhances the ATP triggered activation of Cl channels, likely by binding to a plasma membrane estrogen receptor (Mitchell et al., 2000), but the mode of interaction with the ATP induced signaling cascade is unknown.

ADO

Inflow

FIGURE 8 Purinergic regulation of ciliary epithelial secretion. Following its autocrine release by the NPE cells, ATP is metabolized by ectoenzymes to adenosine, stimulating A3 adenosine receptors to activate Cl channels and enhance inflow. At the stromal surface, ATP released from the PE cells directly stimulates ATP receptors to initiate a cascade leading to activation of maxi Cl channels, thereby reducing net inflow. Tamoxifen synergistically enhances the eVect of ATP.

be developed more fully in Chapter 4 (Macknight and Civan, 2008). The current prevailing view is that water flows from stroma to aqueous humor by local osmosis in response to the osmotic gradient established by the solute transfer.

B. Species Variation

The ionic compositions of the aqueous humor and of the plasma are largely conserved among mammals. One of the largest diVerences reported has been in the HCO3 concentration of the anterior aqueous humor, which is some 28 mM in the rabbit and 22 mM in the human (Krupin and Civan, 1996). The anion gap, defined as the [Naþ concentrationþKþ concentration–Cl concentration] is commonly taken as an approximate index of the HCO3 concentration. Calculated from the data of Gerometta et al. (2005), the anion gap in the aqueous humor of the anterior chamber in several species is 19 mM (sheep), 28 mM (pig and cow), and 40 mM (rabbit). This ranking does not correlate with the calculated values of the anion gap in the plasma of these species. The corresponding anion gap calculated from Table 12 1 of Krupin and Civan (1996) is 45 mM for the rabbit (an overestimate of the measured bicarbonate concentration of 28 mM) and 25 mM for the human (close to the measured value of 22 mM). One interpretation of these measurements would be that there may be a spectrum of bicarbonate concentrations in the aqueous humor, with the sheep at the low end and the rabbit at the high end of the scale. The human bicarbonate concentration is likely close to that of the pig and cow.

With the exception of these relatively minor diVerences, the formation of the aqueous humor largely consists in secreting an isosmotic NaCl solution. It seems reasonable to presume that this secretion is conducted by much the same transporters in diVerent mammalian species. Indeed, bumetanide, Cl channel blockers, and CA inhibitors inhibit transport across the ciliary epithelia of rabbit and cow, and pig as well (Wu et al., 2004; Kong et al., 2006). However, there is increasing awareness of functional diVerences among the several mammalian preparations currently used for experimental study. For example, removing bicarbonate from the bathing solutions qualitatively depolarizes the transepithelial potential (PD) across the isolated ciliary epithelium of several species. However, there are quantitative diVerences. Bicarbonate removal only partially lowers the PD across the ox preparation (Do and To, 2000), abolishes the PD across the pig preparation (Kong et al., 2006), and reverses the PD across the rabbit preparation (Kishida et al., 1981; Krupin et al., 1984). The nonselective Cl channel blocker NPPB is commonly used to block Cl channels in ciliary epithelial cells of other

preparations, but is ineVective in changing PD across the pig ciliary epithelium (Kong et al., 2006). In contrast, another nonselective Cl channel blocker, niflumic acid, also used in studying Cl channels from cells of other species, nearly completely abolishes the PD across the pig preparation (Kong et al., 2006). Whether this pharmacological profile reflects fundamental biophysical diVerences in the porcine channel, or perhaps simply accessibility to the blocking sites, is unknown. In view of these observed diVerences, further study of species variance would be welcome.

C. Circulation

The ciliary plasma flow can be roughly estimated to be 73 ml/min in humans and 50 ml/min in monkeys (Reitsamer and Kiel, 2008). Thus, the maximal diurnal flow of aqueous humor ( 3 ml/min) constitutes only some 4–6% of the plasma flow delivered. As the plasma flow rate falls, the percentage extraction of water from that plasma increases in order to sustain the same rate of aqueous humor secretion. Once the flow rate is reduced by more than 25%, further lowering of plasma flow produces progressive reductions in the rate of aqueous humor formation (Reitsamer and Kiel, 2003, 2008). This phenomenon is analogous to the relationship between the renal plasma flow and glomerular filtration rate (Fig. 33 6D; Giebisch and Windhager, 2005). As in the kidney, the progressive extraction of water necessarily increases the protein concentration of the capillary plasma. This increase in protein concentration elevates the plasma oncotic pressure, restraining further release of water (and with it, solute) from the capillary lumen to the stroma of the ciliary processes. The recent information concerning the dependence of aqueous inflow on circulatory dynamics and its potential significance are considered in Chapter 9 of this volume.

D. Topography

Regional diVerences in the expression of Naþ, Kþ activated ATPase, other proteins and biologically active peptides, summarized by McLaughlin et al. (2001b), led to the suggestion that net ion transport might actually be reversed across some area of the ciliary epithelium (Ghosh et al., 1991). Electron probe X ray microanalyses of intact rabbit ciliary epithelium have provided support for this possibility (McLaughlin et al., 2001b, 2004, 2007). As discussed in Chapter 4 (Macknight and Civan, 2008), this functional topography might provide the basis for a novel approach to reducing net inflow and IOP.