THE YALE JOURNAL OF BIOLOGY AND MEDICINE 50 (1977)

The Role of Paracellular Pathways in Isotonic Fluid Transport* STANLEY G. SCHULTZ University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania Received October 6, 1976 Paracellular pathways across "leaky" epithelia are the major route for transepithelial ionic diffusion. The permselective properties of these pathways suggest that they offer a watery environment through which ions diffuse in their hydrated forms. There is also suggestive evidence that, at least in some tissues, paracellular pathways provide a significant route for transepithelial water flow in response to an osmotic pressure difference; however, this has not as yet been definitively established. The effect ofjunctional complexes that are permeable to ions and water on the predictions of the standing-osmotic gradient model for isotonic water absorption is considered.

CHARACTERISTICS OF THE PARACELLULAR PATHWAY A number of epithelia appear to share a common principal function, namely the transepithelial transport of NaCl from one aqueous milieu to another. Yet, it has been recognized for many years that, in spite of this shared property and many other similarities, the bioelectric characteristics of these epithelia often differ very markedly. For example, so-called "tight" epithelia such as isolated frog skin and toad urinary bladder are characterized by relatively large spontaneous transepithelial electrical potential differences (PDs), when bathed by solutions having identical compositions, and offer relatively large resistances to passive transepithelial ion permeation. In contrast, so-called "leaky" tissues such as small intestine, gallbladder and renal proximal tubule absorb NaCl at rates which generally exceed those of frog skin and toad urinary bladder but are characterized by extremely low and often negligible transepithelial PDs and offer relatively little resistance to transepithelial ionic diffusion. There is now compelling electrophysiologic [1-7] and electron-microscopic [8-10] evidence that these differences cannot be attributed to the possibility that leaky epithelia possess unusually permeable or expansive plasma membranes. Instead, these differences are due to the fact that the zonulae occludens or "tight junctions" of leaky epithelia are relatively highly permeable to ions and thereby open a paracellular ("shunt") pathway for transepithelial ionic diffusion; the conductance of this shunt pathway may account for more than 95 percent [3-6] of the total tissue conductance. Studies by Claude and Goodenough [11] have disclosed morphological differences among junctional complexes from a variety of epithelia which appear to correlate with the degree of "tightness" or "leakiness" so that the anatomic counterpart of the paracellular pathway appears to be firmly established. The relative ionic permeabilities of paracellular pathways across several epithelial tissues are given in Table 1. All of the epithelia listed readily fall into the "leaky" 99 *Supported by research grants from the National Institutes of Health (AM-16275) and the American Heart Association (73-684) Symposium on Isotonic Water Movement, held at the 25th Annual meeting of the American Physiological Society, State University of Albany, August 15, 1974 Please address reprint requests to: Stanley G. Schultz, M.D., Department of Physiology, University of Pittsburgh School, of Medicine, Pittsburgh, PA 15261 Copyright i' 1977 by The Yale Journal of Biology and Medicine, Inc. All rights of reproduction in any form reserved.

100

SCHULTZ 6

os

o:

0.

cra

o

6

V)

6

6

0

0

0

o6

6

~~~~,:0N

II

110

6 CO

Ct

C1 ll

v

6

._

0

6i

vC

0

tf

0

0

'o

00 Cu

.0 (:

0

0

I6oo~

I

cl oo

CO

.-U)

CO

Ho

0

CO.

-.

2

m

0 r-

C0

.0 Cl

-)

C-

-

OCl

Cl

-

Cl

-S

0

-

Cl

-

0

,

_d

.D

;> ._ Iz

04

O

O

)

~~0

_,-~£ -

CO

COH

_L

_E

Ct

O P

Q

O

* 0-

THE SHUNT PATHWAY AND WATER ABSORPTION

101

category inasmuch as they are characterized by low or negligible spontaneous PDs (generally less than 3 mV) and high transepithelial conductances (ranging from 5 mmhos/cm2 for choroid plexus to more than 200 mmhos/cm2 for rat proximal tubule). Several striking similarities are evident. First, the shunt pathways across all of the mammalian epithelia are moderately cation-selective inasmuch as PCI/PK is significantly less than the ratio of the free-solution mobilities of these ions (close to unity). ' Second, selectivity among the alkali cations is minimal to modest; unlike cell membranes, the shunt pathway does not discriminate among these cations by as much as a factor of two. Third, in general, the relative permeabilities of the alkali cations and monovalent anions tend to parallel their relative free solution mobilities so that resistance to permeation tends to increase with increasing hydrated ionic radius. This finding together with the minimal selectivity among the alkali metal ions suggests that the paracellular pathway provides an aqueous environment through which ions diffuse in their hydrated forms [20]. Finally, the shunt pathways across rabbit ileum, rat jejunum and rabbit gallbladder, [13,21], are essentially impermeable to tetraethylammonium ion, a spherical structure with a radius of approximately 4 X; further, rabbit ileum and rat jejunum are essentially impermeable to lysine which has an equivalent radius of approximately 5 X [12,13]. These findings suggest that the limiting dimensions of the shunt pathway restrict the permeation of ions with radii approaching 4-5 X . Moreno and Diamond [21] have estimated the equivalent radii of the paracellular pathways across rabbit and frog gallbladder to be approximately 4-5 X and 6-8 X, respectively.2 As will be discussed below, somewhat larger dimensions have been suggested for human and canine small intestine. It should be stressed that these estimates should not be taken too literally since the configuration(s) of the limiting pathway(s) through the shunt and the non-steric factors that may also influence permeation are unknown. The important point is that the estimated dimensions of paracellular diffusional pathways are such that they should readily permit transepithelial movements of water and small non-electrolytes. THE PARACELLULAR PATHWAY AND ISOTONIC FLUID ABSORPTION One characteristic shared by all "leaky" epithelia, in addition to their low transepithelial PDs and resistances, is that solute transport, be it in the absorptive or secretory direction, is accompanied by that volume of water necessary to render the transported fluid near isotonic.3 The most widely accepted model for this rather precise coupling between solute and water absorption is the "standing-osmotic gradient" model proposed by Diamond and Bossert [22] which is illustrated for the case of absorptive flows in Fig. 1. According to this model, solute (principally NaCl) is extruded from the cell 'PCI!PNa of in vitro rabbit gallbladder appears to increase from very low values (less than 0.08) immediately after dissection to a value of approximately 0.33 after 60-90 min. Barry et al. [ 16] have attributed this to the development, with time, of a second paracellular pathway through which Cl and cations diffuse with relative permeabilities that conform to their relative mobilities in free-solution ("a free-solution shunt"). This explanation clearly cannot apply to the values of PCl/ PNa observed in human jejunum or proximal renal tubule since these tissues were studied in situ. Thus, the properties of the paracellular pathway across rabbit gallbladder appear to depart somewhat from the general properties listed in Table 1. 2Although the fluid absorbed by leaky epithelia is often characterized as isotonic, there is little compelling evidence for exact isotonic absorption in vivo. For example, Soergel et al. [47] have demonstrated that the osmolarity of the fluid absorbed by human jejunum (a leaky epithelium) is approximately 20 percent greater than that of the luminal fluid. As discussed by Schultz and Curran [49], in the presence of an intact circulation transported solutes may be rapidly swept away before complete osmotic equilibration can take place. However, in vitro, transported solutes may be retarded in the subepithelial spaces thereby permitting additional time for osmotic equilibration. 3 The estimates of Moreno and Diamond [21] deal with the equivalent pore radius of the cation-selective pathway since the data used to arrive at these estimates were corrected for diffusion through the "free-solution shunt." Thus, the average overall pore radii of in vitro rabbit and frog gallbladder are undoubtedly larger than the estimates given above.

102

SCHULTZ

(Sb % 0 0 0

~

CAP

%

H

\~~~~~~~~.M.

FIG. I. The "standing osmotic gradient" model for isotonic fluid absorption. The two adjacent cells arejoined together by a "tight junction" which was assumed to be impermeable to ions and water so that all movements are transcellular. BM denotes the basement membrane and CAP an underlying capillary. [Reprinted from Biological Membranes (R. Dowben, editor) with permission from Little, Brown and Company, Inc.]

into the lateral intercellular space which was assumed to be closed at its apical border by a tight junction that is impermeable to solutes and water; this generates a region of local hypertonicity within the space. NaCl then slowly diffuses down the intercellular space and osmotic equilibration takes place by water movement from the cell into this space in response to an osmotic pressure difference. In analyzing this model quantitatively, Diamond and Bossert [22] localized the sites of active Na extrusion from the cell into the lateral intercellular space to the apical one-half of the lateral membranes in order to permit sufficient solute transit-time for complete osmotic equilibration. This point is illustrated in Fig. 2, taken from their paper. The emergent osmolar concentration is given on the ordinate and the fractional length of the interspace from the tight junction at which solute input takes place is given on the abscissa. Two points should be noted. First, as the solute input length increases and approaches the length of the interspace, the emergent fluid becomes increasingly hypertonic. Second, this relation is

103

THE SHUNT PATHWAY AND WATER ABSORPTION

0.38 0.37 o 0.36 0 z 0.35

0 < 0.34

z L 0.33

z

0 (-)

H_

0.32

z

0.31

0.30

_

0

lI

I

50

20

I

30

40

50

60

70

80

90

100

LENGTH OF SOLUTE INPUT (LL) FIG. 2. Computed relation between the final (emergent) osmolarity and the fractional length of the lateral membrane across which solute is transported into the lateral intercellular space. For the upper curve (0) Lp was assumed to be 2 X 10-' cm'/cm2, sec, Osm and for the lower curve (0) it was assumed to be 4X 10-5 cm3/cm2, sec, Osm. Additional details are given in the legend to Fig. 9 of the original paper. [Reprinted from [22] with permission of the Rockefeller University Press.]

extremely sensitive to the hydraulic conductivity (Lp). A two-fold increase in Lp from the upper curve (open circles) to the lower curve (closed circles) markedly extends the permissible length of solute input consistent with an isotonic or near isotonic absorbate. Thus, according to this model, the osmotic pressure difference is greatest in the apical portion of the cul-de-sac and is gradually dissipated by water flow across the lateral membranes so that the emergent fluid is (near) isotonic; hence, the name

"standing-osmotic gradient." In the opinion of this author, the weakest consequence of this ingenious model is the need to localize active solute transport mechanisms to a specific region of a continuous and presumably "fluid" membrane that otherwise appears to be homogeneous. Although considerable evidence has accrued that the Na-K stimulated, ouabain-inhibited ATPase which may be involved in active Na transport by epithelial cells is found in the baso-lateral membranes (and only to a much lesser degree, if at all, in the mucosal membranes) [23-27], there is no evidence that this activity is restricted to the apical portions of the lateral membranes. The autoradiographic studies of Stirling [27] employing 3H-ouabain are consistent with the notion that almost all of this ATPase in rabbit ileum is localized in the baso-lateral membranes and that little or none is found in the brush border; however, the result of these studies suggest a uniform distribution

SCHULTZ

104 of

activity along

conclusive,

the

fact

the

Although this evidence

membranes.

lateral

remains

that

date

to

no

direct evidence for

cannot

a

be

considered

localization

of pump

sites along the baso-lateral membranes has been forwarded. Thus, the standing osmotic gradient model for isotonic water absorption was founded on an assumption which, though entirely reasonable at the time, has proven to be

at

least

model

incorrect.

partially

remains

established,

unsupported. may inquire:

we

At

the

same

Now that

time,

How would the

principal

of the

one

paracellular

opening

predictions of this

for ion transport

routes

of the

tight junctions

firmly flow

are

the

to

of ions and water, as illustrated in Fig. 3, alter the consequences and predictions of the original standing osmotic gradient model? And, in particular, to what extent would sites along the need localize baso-lateral

this

modification

relieve

the

the

to

pump

membrane? question

A

that

is

central

contribute significantly to relative

contributions

response

to

region

a

of

of

inquiries

these

to

is:

transcellular

and

more

specifically, what

paracellular pathways

hypertonicity established

pathway

the

osmotic water flow and,

the

paracellular

Does

to

flow in

intercellular

within the lateral

the

are

water

space?

THE CONTRIBUTION OF THE PARACELLULAR PATHWAY TO OSMOTIC WATER FLOW Currently, the relative to

water

pressure

across

flow

is

difference

an

hydraulic conductivity through

shunt

in response

to

(Lp),

across

and

other

spontaneous

a

unsettled and controversial

pathway across

that the shunt the

contributions of the transcellular and paracellular pathways

leaky epithelia

issue.

or

There is

induced

suggestive

osmotic evidence

tissues strongly influences the transepithelial

some

equally suggestive evidence tissues

is

minimal.

We

will

that

osmotic water

consider

each

flow of

body

evidence in turn. Evidence that the shunt pathway contributes significantly to As mentioned above, one physiologic characteristic of leaky epithelia is that all are capable of bringing about the movement of relatively large volumes of near isotonic fluids; according to the standing-osmotic gradient model, this ability must be closely linked to the hydraulic conductivities of these tissues. Fig. 4 is a plot of the transepithelial

Lp:

conductances

(G, in mmhos/CM2)

versus

(Lp, in

the hydraulic conductivities

CM3/CM2,sec,Osm) of 14 epithelial tissues;4 the line shown has a slope of unity. Clearly, on the average, an increase in G between the tight epithelia, clustered on the lower left, and the leaky epithelia, in the upper right, is associated with a nearly proportional increase

in

Lp.'

If we

to

ions,

it

seems

likely

however,

to

intended

simply

'The

plotted

described

highly likely

many

clamped

high conductances of "leaky"

escape the

to

epithelia

conclusion that the shunt

pathway

by a

G

Lp and! or

to illustrate the

4

direct relation between

G

many

water

by

and

Lp of different epithelia; the fact

expect an increase in

that the

reasons. First,

"edge damage" inflicted when the

The effect of edge damage on

4p particularly

in

4p has

not been

examined

"tight" epithelia. The values for 4p are likely tissue.

and dilute solutes in the

transepithelial osmotic pressure

unstirred

layer facing the solution toward which water

difference,

this

"concentrating

and

washing-out"

to be

It has been

water flow across a membrane will tend to concentrate solutes in the unstirred layer

flows

it

epithelial

introduced by edge damage is certainly likely to be considerably

presence of significant unstirred layers adjacent to the two surfaces of the

years that

solutionfrom

G

are overestimates due to

two

might

epithelia

shown. Further, in some instances different investigafor the same tissue; the values chosen for Fig. are considered

straight line with a slope of unity should not be overstressed for several

half-chambers; the error "tight" epithelia than for "leaky"

intuitively,

is

significantly affect the pattern

instances the values of

the

flow

that the

epithelia.

greater

recognized

difficult

different values for

reported

representative.

directly

premise

nonmammalian epithelia were obtained at room temperature and those for mammalian

4In general,

370

the

of the presence of water-filled paracellular channels that are highly

are consequences permeable

accept

facing

flows.

effect

will

THE SHUNT PATHWAY AND WATER ABSORPTION

H20

105

No

FIG. 3. Modified version of the standing osmotic gradient model which permits paracellular movements of Na and water. decrease the initial difference in osmolarity and the steady-state rate of water flow. Thus, Lp's calculated from the observed water flow divided by the presumed osmolarity difference, will be underestimates (i.e., flow will be reduced in comparison with the osmolarity difference between the bulk solutions). Dainty[50] has shown that when water flows across a membrane from compartment I to compartment 2, the concentration of solute i in the unstirred layer of compartment 1 will be given by

Cu(,> = Cb(,) exp (v6(l)/DI) and the concentration of solute i in the unstirred layer of compartment 2 is given by

Cu(2)

=

Cb(2) exp (- v6(2)/ Di)

where the subscripts u and b designate the unstirred layer and bulk solution respectively, v is the linear velocity of water flow, 6 is the thickness of the unstirred layer and Di is the diffusion coefficient of i in water. Clearly, the extent to which unstirred layers will lead to underestimates of Lp increases with both the thickness of the layers (6(1) + 6(2)) and the linear velocity of water flow (v). Assuming that, under optimal stirring conditions, the thickness of the unstirred layers surrounding epithelia do not differ markedly, the major determinant of the error due to the presence of unstirred layers is the rate of water flow. Thus, the greater the "true" Lp, the greater this value will be underestimated from steady-state measurements. In short, all other parameters being equal, the true Lp's ofleaky epithelia will be underestimated to a greater degree than those of tight epithelia. For example, Dainty and House [39] have reported that the Lp for isolated frog skin is unaffected by increasing the stirring rate from 0 to 500 revolutions per minute; thus, the velocity of water flow across this tight epithelium is sufficiently low so that the "concentrating and washing-out" effects in the unstirred layers are negligible. In contrast, Diamond [50] reported that the Lp of rabbit gallbladder is 5 X 10- cm3/cm2, sec,Osm; this value was increased by one order of magnitude by the studies of Wright et al. [321] and van Os, using a technique which permits rapid measurements of water flow (i.e., within 30 sec. of applying a pressure difference) so that concentrating and washing-out effects are reduced, reported an 4 for rabbit gallbladder of 4 X 0-:3 cm3/cm2, sec, Osm [51]. Clearly, then, the steady-state Lp's of highly leaky epithelia are underestimates and may be in error by more than two orders of magnitude. Attempts to correct these values using the equations given above are not entirely satisfactory because the calculated value of v strongly depends upon the presumed pathways for volume flow and the fractional area occupied by these pathways. Methods that permit the determination of the "initial" rate of water flow in response to a hydrostatic or osmotic pressure difference would circumvent many of these problems but they pose formidable experimental difficulties.

106

SCHULTZ

200 _

,

100

_

E 10

_,

//

I. 0

-~~~~2

44

/

0

104

*8~~*

0.1

l

lo-6

io-5

I

l

io-4

io-3

I

_,

io-2

L p (c m3/CM2 sec Psm) FIG. 4. Lp vs. G for a number of tight and leaky epithelia. I = rat proximal tubule [28,29]; 2 =rat jejunum [13,30]; 3 = rabbit gallbladder [31,32]; 4 = fish gallbladder [33,34]; 5 = Necturus gallbladder [4,5]; 6 = frog choroid plexus [5,18,35]; 7 =frog stomach [36,37]; 8 = frog skin [38-40]; 9 = toad urinary bladder [41,42]; 10 = turtle urinary bladder [43,44]; I I Necturus proximal tubule [6,45]; 12 = human ileum [46,47]; 13 = human jejunum [47,48]; 14 = rat distal tubule [29].

also responsible for the high Lp. Alternatively, it is conceivable that epithelia with high conductance shunt pathways also possess unusually leaky plasma membranes which are responsible for the relation shown in Fig. 4. The data bearing on this point are limited because of the difficulty of determining membrane resistances in non-planar epithelia. However, Table 2 gives the resistances of the transcellular and paracellular pathways and the hydraulic conductivities of two leaky and two tight epithelia. Clearly', where determined, the resistances of the two limiting cell membranes and the resistance of the transcellular pathway, (Rm, + RS), of tight and leaky epithelia do not differ markedly. However, the resistances of the paracellular pathways, (RL), differ by more than one order of magnitude and tend to parallel the differences in LP. Although the data in Fig. 4 and Table 2 are not conclusive, they certainly suggest that a single anatomic feature of leaky epithelia is responsible for both the leakiness to current flow and the leakiness to osmotic water flow.6 In other words, these data suggest that the paracellular shunt pathway is not only the major route for transepithelial ionic 6A priori, there is no reason to conclude that the resistance to current flow must parallel the resistance to osmotic water flow. However, the results of studies on artificial lipid membranes treated with "pore-forming" antibiotics indicate that there is a linear relation between increases in electrical conductivity and increases in Lp and the diffusional permeabilities of nonelectrolytes [52]. Thus, it is not unreasonable to suspect that membranes having similar electrical resistances also have similar Lp's.

THE SHUNT PATHWAY AND WATER ABSORPTION

107

TABLE 2 Transcellular and Paracellular Resistances and Hydraulic Conductivities of "Tight" and "Leaky" Epitheliat

Rm

Rs

(Rm+Rs)

RL

ohm cm2

"Tight" Epithelia Frog skin [38] Toad urinary bladder [41]

LP ml/ cm2,sec,Osm

-

-

2800

3800

3600

7400

24,000 12,320

9 X 10-6 [6,27] 7 X 10-6 [23]

"Leaky" Epithelia Necturus gallbladder [4] 4500 2900 7400 320 I X l0-4 * 1 X 10-4 [2] -6000 Necturus renal proximal -2000 7900 70 tubule [6] *The Lp of Necturus gallbladder has not been established; the value given is that for fish gallbladder [8] which is likely to be a reasonable estimate. t R. and Rs are the apparent resistances of the mucosal and serosal membranes, respectively; (R, + R,) is the resistance of the transcellular pathway; and RL is the apparent resistance of the paracellular pathway.

diffusion but also is a major determinant of the transepithelial hydraulic conductivity. The results of studies on the relation between water flow and solute flow across human and canine small intestine also suggest that the shunt pathway provides a major route for water flow. In human jejunum, transepithelial movements of Na, K and urea are markedly affected by bulk flow; the apparent reflection coefficients for these solutes are only 0.5, 0.4 and 0.5, respectively [53-55].7 The data obtained by Fordtran et al. [53,54] and Soergel et al. [47] are consistent with the notion that the principal pathway for spontaneous and osmotically induced water flows across human jejunum is characterized by an "equivalent pore radius" of 7-9 X, a value that is considerably larger than that estimated for most cell membranes but well within the range of values estimated for the dimensions of shunt pathways. The studies of Lifson and his collaborators [56,57] on the diffusive and convective components of solute flow across in vivo canine jejunum permit a similar conclusion. These investigators found a linear relation between the rate of spontaneous water absorption and the absorption of urea, arabinose, xylose and glucose in the absence of a significant transepithelial concentration difference for these solutes.8 The apparent reflection coefficients, calculated from the ratio of the concentration of the solute in the absorbate and that in the luminal fluid, were 0.2 for urea and approximately 0.4 for arabinose, xylose and glucose. Thus, there was relatively little "sieving" of molecules with equivalent radii up to at least 4 A and the data are consistent with the presence of a transepithelial pathway with an "equivalent pore radius" of approximately 15 X. On the other hand, the diffusive components of xylose and arabinose transport (estimated from transport driven by a concentration difference in the absence of net water movement) were small and suggested that the 'The apparent reflection coefficients for Na and urea determined from experiments examining the "osmotic effectiveness" of these solutes [53] and from experiments determining the effect of water flow on the flows of these solutes [54] yielded essentially identical results. 8As discussed in footnote 5 and reference [58], as the result of unstirred layers, water flow will necessarily establish transepithelial concentration gradients even when the concentrations in the bulk solutions are identical. Thus, a direct relation between water flow and solute flow need not mean that there is a coupled interaction between water and solute flows through a common pathway (i.e., "solvent-drag"). Water flow may simply establish a concentration difference between the unstirred layers which provides the driving force for solute diffusion through pathways other than those traversed by water. However, the equations given in footnote 5 indicate that if solute flow is simply the result of a "concentrating and washingout effect," the relation between water flow and solute flow should be non-linear. The linear relations observed by Fordtran et al. [54] and Lifson et al. [56,57] strongly suggest direct interactions between water and solute flow through common channels.

108

SCHULTZ

diffusional pathway has an equivalent pore radius of only 5 X. This discrepancy between the estimated dimensions of the convective pathway and the diffusional pathway strongly suggests that the epithelium must be viewed, to a first approximation, as a mosaic membrane comprising at least two different parallel pathways. Diffusion is permitted to take place through both pathways and the permeabilities of these pathways are such that the area-weighted average is consistent with an equivalent pore radius of 5 A. However, if water flow predominantly takes place through the larger but far less numerous pathway, pore radii calculated from apparent reflection coefficients will approach the dimensions of this route.9 In the present context, it is not unreasonable to identify one pathway with the transcellular route and the other with the paracellular route and to argue that the latter is the principal pathway for spontaneous or osmotically induced water flow across canine jejunum and has an equivalent radius in the range of 10-15 A. Fromter et al. [60] have analyzed Na and Cl transport by rat renal proximal tubule in terms of the phenomenologic equations of irreversible thermodynamics. They report that the reflection coefficients of Na and Cl are 0.7 and 0.5, respectively, and that onethird of net Na reabsorption and one-half of net Cl reabsorption are attributable to solvent-drag. These estimates are in excellent agreement with those reported by Kokko et al. [see 29] for rabbit proximal tubule but are somewhat larger than the reflection coefficient for NaCl of 0.4 reported by Rector et al. [see 29] for the rat. In any event, the reflection coefficients of Na and Cl are consistent with a solvent-drag pathway having a pore radius of 6 X if one uses the (uncorrected) Stokes-Einstein hydrated ionic radii and 8-10 X if the corrected Stokes-Einstein radii are employed. These data further imply that more than 60 percent of total water absorption across rat proximal tubule takes place through this solvent-drag pathway. Thus, it appears that for small intestine and proximal tubule, the paracellular pathway is not only a major route for transepithelial ionic diffusion but also strongly influences, if not dominates, coupled interactions between the transepithelial flows of solutes and water and, therefore, must provide a significant route for transepithelial water movement. Evidence that the paracellular pathway contributes minimally to Lp: To date, evidence against the notion that water flow through the paracellular pathway contributes significantly to the overall Lp stems mainly from studies on rabbit gallbladder. Wright et al. [32] calculated the total area occupied by pores through the tight junctions from measurements of transepithelial electrical resistance. Then, assuming that the pores have an equivalent radius of 12 X [44] and that water flow through these channels conforms to Poiseuille's law, they calculated that the toal Lp of the paracellular route is approximately 1 X 10-5 cm3/cm2,sec,Osm, a value that is only 10 percent of the overall transepithelial Lp. 9Kedem and Katchalsky [59] have shown that if a membrane is traversed by two different parallel pathways, a and,, one occupying a fractional areaYa and the other occupying a fractional area y , then:

LP= yLp +;LP u=

W. +Y wo

and a=

LP

a +

(y,3LPf /L,p)au

where wRRTis the overall conventional permeability coefficient (P); a is the overall reflection coefficient; the subscripts a and ,O designate the two pathways; andya, + -y = 1. Then, ifY > -Y butLp,8> Lpa, it is possible thatc.v w,but a cf. Thus, the overall permeability coefficient (ocRT) will primarily reflect the permeability of the more expansive pathway but the reflection coefficient will be determined primarily by the predominant pathway for osmotic water flow.

109

THE SHUNT PATHWAY AND WATER ABSORPTION

van Os arrived at a similar conclusion [51] using entirely different data. This investigator applied

equation [61]

Renkin

the

permeabilities

the diffusive

to

rabbit

of

estimated that the limiting pore radius pathway is 40 X. Then, employing Poiseuille's law, he calculated the Lp attributable to the paracellular pathway is 8 X 10-5 cm3/cm2,sec,Osm (per cm2 tissue) compared to an overall transepithelial Lp of approximately 4 X l0o3 thus, according to these calculations, approximately 2 percent of total transepithelial water flow in response to an osmotic pressure difference circumvents the transcellular route. It is of interest that Wright et al. and van Os arrived at similar conclusions in spite of the fact that the equivalent pore radius estimated by

gallbladder to a variety of non-electrolytes and of the paracellular

that

cm3/cm2,sec,Osm; Wright

al.

et

(12 X)

one-third

is

by

is that theLp estimated by

apparent coincidence

al.

by Wright

estimated

that estimated

Wright

Had

et

al.

Os

van

X);

(40

the

this

for

reason

Os is 40 times greater than that

van

employed

radius

a

X

of 40

their

in

Lp attributable

to the paracellular pathway would have readily accounted for the entire transepithelialLp.

calculations, the

important point

The

is

that

direct

the

inasmuch

as

interactions

strongly

it

about

known

the

probe

permeation, totality

be

interpreted

Moreno

and

estimates

of

with

Diamond

the

of

the

molecules

are

not

factors

of

on

radius,

equation

diffusion

and

the

taken into

that

parameter that is

a

Renkin

subject to

not

subject

to

steric;

entirely

are

question

non-steric

channels, which consideration. Until much paracellular

may

more

is

and

permeation of charged

influence

may

is

estimates of pore size and calculations based on these estimates

uncharged molecules, must

of the

restrictions

that

between

influence

application

The

assumes

contribution

calculate the relative

value of the pore

assumed

measurement.

to

osmotic water flow employing Poiseuille's law is critically

paracellular pathway to dependent upon

any attempt

pore

great deal of

[21],3

radius

Wright across

caution. This point is illustrated al. [32,62], [51] and

et

rabbit

gallbladder

Os

van

that

by the fact

have

vary

over

a

that

arrived at

near

tenfold

range.

CONCLUSIONS AND IMPLICATIONS There

is

no

doubt

that

leaky epithelia provide mucosal

solution

pathway

often

which

these

pathways

important

and

lateral

markedly

that

traverse

extracellular

tight junctions

the

route

of

number

a

for ionic diffusion

intercellular spaces and that the conductance

exceeds that of the transcellular

paracellular pathways

contribute

to

route. However,

transepithelial

osmotic

of

this

the extent water

of

the

between

to

flow is

an unsettled question. Nevertheless, it is of interest to consider the implications of tight junctional complexes that are leaky to both ions and water with respect to the original

standing-osmotic gradient model for isotonic water absorption. In particular, we may inquire (a) the

lateral

how

would

leaky junctions affect

such

intercellular

the osmotic pressure

space and (b) whether the need

to

profile

localize active solute

within pumps

at the apical region of the lateral membranes would be relieved? Clearly, opening diffusion

of

the

transported

tight junctions

illustrated

solutes, would by itself

(i.e.,

in

Fig. 1

so

assuming

to

as

that the

permit back-

transjunctional

pathway is not a major route for osmotic water flow) alter the concentration or osmotic pressure profile maximum

Further, if, and

water,

junctional

within

would

be

as illustrated

the

end

the

in

some

Fig. 3,

concentration

of the

interspace.

displaced

In

short,

directed

these

conditions, apical

junctional complexes

maximum would be

interspace.

complexes permit oppositely

the

Under

distance from the

net

it

is

even

are

further

end

of

displaced

water

concentration

interspace.

of the

permeable

intuitively obvious

movements

the

that

if

to

both

ions

away from

the

and solutes

the

junctional

to

and

from

110

SCHULTZ

the intercellular space, a "standing-osmotic gradient" with a maximum at the apical end of the interspace would no longer prevail. Further, the need to localize the sites of active solute transport into the interspace to the apical region stemmed from the assumption that osmotic equilibration could only result from transcellular water flow and that a sufficiently long transit time was necessary in order to complete this process. If the junctional complexes permit osmotically driven water flow at a sufficiently rapid rate, under steady-state conditions the osmolarity of the fluid in the apical region of the interspace will approach that of the mucosal solution so that the sites of active solute input consistent with a near isotonic absorbate may be extended further along the lateral membranes. The paper by Boulpaep and Sackin presents a quantitative treatment of this problem; suffice it to say at this point that the opening of the junctional complexes to the diffusion of ions and the osmotic flow of water must augment the ability of leaky epithelia to elaborate an isotonic absorbate and may entirely relieve the necessity for localizing active solute pumps to the apical regions of the interspace in future attempts to model this process. ADDENDUM This manuscript was completed in September 1974. Since then, a number of observations have been reported that reinforce the notion that paracellular pathways across "leaky" epithelia provide a significant route for transepithelial nonelectrolyte and isotonic water transport. The studies of Berry and Boulpaep [63] on the paracellular movements of sucrose across Necturus proximal tubule suggest that the "slit width" of the junctional pathway is approximately 14 X and that the apparent reflection coefficient for this solute is only 0.7. Schafer et al. [64] and Sackin and Boulpaep [65] have analyzed salt and water absorption by isolated segments of rabbit pars recta and Necturus proximal tubule, respectively, using a modification of the Diamond-Bossert approach but assuming that the paracellular pathway is permeable t6 ions and water. Their results suggest that (near) isotonic fluid absorption can be accomplished in the absence of a significant standing osmotic gradient in the intercellular spaces. Huss and Marsh [66] have presented a model of salt and water transport across renal proximal tubule, that also assumes osmotic coupling between solute and water flow, in which the importance of determining the contribution of the paracellular water flow is clearly outlined. Finally, Hill [67] has raised the objection that the values chosen by Diamond and Bossert [22] for the length (L = 100,u) and radius (r = 0.05,u) of the intercellular space [see Figs. 3,4,7,8,9 and 10 of ref.22] are unrealistic, and concluded that using more "realistic" dimensions, " . it seems virtually impossible that the intracellular and lateral spaces of fluid-transporting epithelia are in fact osmotic coupling spaces." More recently, Hill [68] reported that when the lumen of Necturus gallbladder is exposed to solutions of NaCl having osmolarities as low as 5 mOsm, the osmolarity of the absorbate equals that of the luminal solution. These observations, if correct, cannot be readily reconciled with the notion that isotonic water transport is the result of osmotic equilibration of water between the cells and hypertonic regions in the intercellular spaces, and, instead, would strongly suggest a paracellular route for water flow secondary to solute transport. Thus at present, the details of the mechanism(s) responsible for isotonic water

THE SHUNT PATHWAY AND WATER ABSORPTION

111

transport by "leaky" epithelia and, in particular, the role of paracellular pathways are unclear. A more precise knowledge of the dimensions and hydraulic conductivities of these "shunts" is certainly necessary before these important physiologic questions can be resolved. REFERENCES 1. Rose RC, Schultz SG: Studies on the electrical potential profile across rabbit ileum: Effects of sugars and amino acids on transmural and transmucosal P.D.'s. J Gen Physiol 57:639-663, 1971 2. Boulpaep EL, Seely JF: Electrophysiology of proximal and distal tubules in the autoperfused dog kidney. Am J Physiol 221:1084-1096, 1971 3. Frizzell RA, Schultz SG: Ionic conductances of extracellular shunt pathway in rabbit ileum: Influence of shunt on transmural sodium transport and electrical potential differences. J Gen Physiol 59:318-346, 1972 4. Fr.smter E: The route of passive ion movement through the epithelium of Necturus gallbladder. J Membrane Biol 8:259-301, 1972 5. Fromter E, Diamond J: Route of passive ion permeation in epithelia. Nature (New Biology) 235:9-13, 1972 6. Windhager EE, Boulpaep EL, Giebisch G: Electrophysical studies on single nephrons. Proc Intern Congr Nephrol, 3rd. Washington, DC, 1966, vol. I, pp 35-47 7. Reuss L, Finn AL: Electrical properties of the cellular transepithelial pathway in Necturus gallbladder. J Membrane Biol 25:115-139, 1975 8. Machen TE, Erlij D, Wooding FBP: Permeable junctional complexes: The movement of lanthanum across rabbit gallbladder and intestine. J Cell Biol 54:302-312, 1972 9. Whittembury G, Rawlins FA: Evidence of a paracellular pathway for ion flow in the kidney proximal tubule: Electromicroscopic demonstration of lanthanum precipitate in the tight junction. Pflug Arch 330:302-309, 1971 10. Tisher CC, Yarger WE: Lanthanum permeability of the tight junction (zonula occludens) in the renal tubule of the rat. Kidney Internat 3:238-250, 1973 11. Claude P, Goodenough DA: Fracture faces of zonulae occludentes from "tight" and "leaky" epithelia. J Cell Biol 58:390-400, 1973 12. Munck BG, Schultz SG: Lysine transport across isolated rabbit ileum. J Gen Physiol 53:157-182, 1969 13. Munck BG, Schultz SG: Properties of the passive conductance pathway across in vitro rat jejunum. J Membrane Biol 16:163-174, 1974 14. Wright EM: Diffusion potentials across the small intestine. Nature (London) 212:189-190, 1966 15. Turnberg LA, Bieberdorf FA, Morawski SG, et al: Interrelationships of chloride, bicarbonate, sodium and hydrogen transport in human ileum. J Clin Invest 49:557-567, 1970 16. Barry PH, Diamond JM, Wright EM: The mechanism of cation permeation in rabbit gallbladder. J Membrane Biol 4:358-394, 1971 17. Frlmter E, Muller CW, Wick T: Permeability properties of the proximal tubular epithelium of the rat studied with electrophysiological methods, Electrophysiology of Epithelial Cells, Stuttgart, FK Schattauer Verlag, 1971, pp 119-146 18. Wright EM: Mechanisms of ion transport across the choroid plexus. J Physiol (London) 226:545-571, 1972 19. Robinson RA, Stokes RH: Electrolyte solutions. 2nd Edition. New York, Academic Press, 1959 20. Eisenman G: On the elementary atomic origin of equilibrium ionic specificity, Membrane Transport and Metabolism. Edited by A Kleinzeller and A Kotyk. New York, Academic Press, 1961, pp 163-179 21. Moreno JH, Diamond JM: Cation permeation mechanisms and cation selectivity in "tight junctions" of gallbladder epithelium. Membranes-A series of Advances. Edited by G Eisenman. New York, Marcel Dekker, Inc., 1974 22. Diamond JM, Bossert WH: Standing-gradient osmotic flow: A mechanism for coupling of water and solute transport in epithelia. J Gen Physiol 50:2061-2083, 1967 23. Fujita M, Ohta H, Kawai K, et al: Differential isolation of microvillous and baso-lateral plasma membranes from intestinal mucosa: Mutually exclusive distribution of digestive enzymes and ouabain-sensitive ATPase. Biochim Biophys Acta 274:336-347, 1972 24. Fujita M, Matsui H, Nagano K, et al: Asymmetric distribution of ouabain-sensitive ATPase in rat intestinal mucosa. Biochim Biophys Acta 233: 404-408, 1971 25. Quigley JP, Gotterer GS: Distribution of (Na + K)-stimulated ATPase activity in rat intestinal mucosa. Biochim Biophys Acta 173:456-468, 1969 26. Schmidt U, Dubach UC: Na-K stimulated adenosinetriphosphatase: Intracellular localization within the proximal tubule of the rat nephron. Pflugers Arch 330:265-270, 1971 27. Stirling CE: Radioautographic localization of sodium pump sites in rabbit intestine. J Cell Biol 53:704-714, 1972 28. Hegel U, Frbmter E, Wick T: Der elektrische Wandwiderstand des proximalen Konvolutes der Rattenniere. Arch Ges Physiol 294:274-290, 1967

112

SCHULTZ

29. Ullrich KJ: Permeability characteristics of the mammalian nephron. Handbook of Physiology, Section 8: Renal Physiology. Edited by J Orloff and RW Berliner, Washington, DC, American Physiological Society, 1973, pp 377-398 30. Smyth DH, Wright EM: Streaming potentials in the rat small intestine. J Physiol (London) 182:591-602, 1966 31. Wright EM, Barry PH, Diamond JM: The mechanism of cation permeation in rabbit gallbladder: Conductances, the current-voltage relation, concentration dependence of anion-cation discrimination and the calcium competition effect. J Membrane Biol 4:331-357, 1971 32. Wright EM, Smulders AP, Tormey JMcD: The role of the lateral intercellular spaces and solute polarization effects in the passive flow of water across rabbit gallbladder. J Membrane Biol 7:198-219, 1972 33. Diamond JM: The mechanism of solute transport by the gallbladder. J Physiol (London) 161:474-502, 1962 34. Diamond JM: The mechanism of water transport by the gallbladder. J Physiol (London) 161:503-528, 1962 35. Wright EM: Ion transport across the frog posterior choroid plexus. Brain Res 23:302-304, 1970 36. Hogben CAM: Active transport of chloride by isolated frog gastric epithelium. Origin of the gastric mucosal potential. Am J Physiol 180:641-649, 1955 37. Durbin RP, Frank H, Solomon AK: Water flow through frog gastric mucosa. J Gen Physiol 39:535-551, 1956 38. Ussing H H, Windhager EE: Nature of shunt path and active sodium transport path through frog skin epithelium. Acta Physiol Scand 61:484-504, 1964 39. Dainty J, House CR: An examination of the evidence for membrane pores in frog skin. J Physiol (London) 185:172-184, 1966 40. House CR: The nature of water transport across frog skin. Biophys J 4:401-416, 1964 41. Reuss L, Finn AL: Passive electrical properties of toad urinary bladder epithelium: Intercellularelectrical coupling and transepithelial cellular and shunt conductances. J Gen Physiol 64:1-25, 1974 42. Hays RM, Leaf A: Studies on the movement of water through the isolated toad bladder and its modification by vasopressin. J Gen Physiol 45:905-919, 1962 43. Solinger RE, Gonzolez CF, Shamoo YE, et al: Effect of ouabain on ion transport mechanisms in the isolated turtle bladder. Am J Physiol 215:249-260, 1968 44. Schilb TP, Brodsky WA: Transient acceleration of transmural water flow by inhibition of sodium transport in turtle bladders. Am J Physiol 219:590-596, 1970 45. Bentzel CJ, Parsa B, Hare DK: Osmotic flow across proximal tubule of Necturus. Correlation of physiologic and anatomic studies. Am J Physiol 217:570-580, 1969 46. Grady GF, Madoff MA, Duhamel RC, et al: Sodium transport by human ileum in vitro and its response to cholera enterotoxin. Gastroent 53:737-744, 1967 47. Soergel KH, Whalen GE, Harris JA: Passive movement of water and sodium across human small intestinal mucosa. J Appl Physiol 24:40-48, 1968 48. Rohde JE, Andersen B: In vitro measurement of ion fluxes across biopsies of human jejunal mucosal during cholera. J Appl Physiol 35:557-561, 1973 49. Schultz SG, Curran PF: Intestinal absorption of sodium chloride and water, Handbook of Physiology. Alimentary Canal. Washington, DC, Am. Physiol. Soc., 1968, sect. 6, vol. III, pp 1245-1275 50. Diamond JM: The mechanism of isotonic water transport. J Gen Physiol 48:15-42, 1964 51. van Os CH: Transport parameters of isolated gallbladder epithelium. Doctoral Thesis: University of Nijmegen, Nijmegen, The Netherlands, 1974 52. Holz R, Finkelstein A: Water and nonelectrolyte permeability induced in thin lipid membranes by the polyene antibiotics Nystatin and Amphotericin B. J Gen Physiol 56:125-145, 1970 53. Fordtran JS, Rector FC Jr, Ewton MF, et al: Permeability characteristics of the human small intestine. J Clin Invest 44:1935-1944, 1965 54. Fordtran JS, Rector FC Jr, Carter NW: The mechanisms of sodium absorption in the human small intestine. J Clin Invest 47:884-900, 1968 55. Turnberg LA: Potassium transport in the human small bowel. Gut 12:811-818, 1971 56. Levitt DG, Hakim AA, Lifson N: Evaluation of components of transport of sugars by dog jejunum in vivo. Am J Physiol 217:777-783, 1969 57. Lifson N, Hakim AA: Simple diffusive-convective model for intestinal absorption of a nonelectrolyte (urea). Am J Physiol 211:1137-1146, 1966 58. Schultz SG: Irreversible Thermodynamics, Biomembranes, Volume 4A: Intestinal Absorption. Edited by DH Smyth. London, Plenum Press, 1974, pp 199-239 59. Kedem 0, Katchalsky A: Permeability of composite membranes: Part 2. Parallel elements. Trans Faraday Soc 488:1931-1940, 1963 60. Fromter E, Rumrich G, Ullrich KJ: Phenomenologic description of Na, Cl and HCO3 absorption from proximal tubules of the rat kidney. Pflugers Arch 343:189-220, 1973 61. Renkin EM: Filtration, diffusion and molecular sieving through porous cellulose membranes. J Gen Physiol 38:225-243, 1955

THE SHUNT PATHWAY AND WATER ABSORPTION

113

62. Smulders AP, Wright EM: The magnitude of nonelectrolyte selectivity in the gallbladder epithelium. J Membrane Biol 5:297-318, 1971 63. Berry CA, Boulpaep EL: Nonelectrolyte permeability of the paracellular pathway in Necturus proximal tubule. Am J Physiol 228:581-595, 1975 64. Schafer JA, Patlak CS, Andreoli TE: A component of fluid absorption linked to passive ion flows in the superficialpars recta. J Gen Physiol 66:445-471, 1975 65. Sackin H, Boulpaep EL: Models for coupling of salt and water transport. Proximal tubular reabsorption in Necturus kidney. J Gen Physiol 66:671-733, 1975 66. Huss RE, Marsh DJ: A model of NaCl and water flow through paracellular pathways of renal proximal tubules. J Membrane Biol 23:305-347, 1975 67. Hill AE: Solute-solvent coupling in epithelia: a critical examination of the standing-gradient osmotic flow theory. Proc R Soc Lond B 190:99-114, 1975 68. Hill AE: Fluid transport by the gall-bladder epithelium of Necturus. J Physiol (London) 263:201P, 1976

Stanley G. Schultz, M.D. Department of Physiology University of Pittsburgh School of Medicine Pittsburgh, PA 15261