The microcirculation of the renal
medulla traps NaCl and urea deposited to the interstitium by the loops
of Henle and collecting ducts. Theories have predicted that
countercurrent exchanger efficiency is favored by high permeability to
solute. In contrast to the conceptualization of vasa recta as simple
"U-tube" diffusive exchangers, many findings have revealed
surprising complexity. Tubular-vascular relationships in the outer and
inner medulla differ markedly. The wall structure and transport
properties of descending vasa recta (DVR) and ascending vasa recta
(AVR) are very different. The recent discoveries of aquaporin-1 (AQP1)
water channels and the facilitated urea carrier UTB in DVR endothelia
show that transcellular as well as paracellular pathways are involved
in equilibration of DVR plasma with the interstitium. Efflux of water
across AQP1 excludes NaCl and urea, leading to the conclusion that both
water abstraction and diffusion contribute to transmural equilibration. Recent theory predicts that loss of water from DVR to the interstitium favors optimization of urinary concentration by shunting water to AVR,
secondarily lowering blood flow to the inner medulla. Finally, DVR are
vasoactive, arteriolar microvessels that are anatomically positioned to
regulate total and regional blood flow to the outer and inner medulla.
In this review, we provide historical perspective, describe the current
state of knowledge, and suggest areas that are in need of further exploration.
vasa recta; microperfusion; microcirculation; water channel; urinary concentration; permeability
 |
INTRODUCTION |
SINCE THE
EXPERIMENTAL FINDINGS of Wirz et al. (147) led to
the countercurrent theory of the urinary concentrating mechanism, as
described by Hargitay and Kuhn (31), most subsequent
research has focused on the countercurrent multiplier function of the
loops of Henle. According to the theory, a small difference in osmotic pressure (the single effect) is multiplied by countercurrent flow in
adjacent channels of the limbs of Henle's loop to produce a large
axial difference in osmotic pressure between the renal cortex and the
tip of the renal papilla; that is, the multiplier generates a
hypertonic renal medulla. Less attention has been paid to
countercurrent exchange, which is thought to preserve medullary
hypertonicity rather than create it. It is generally accepted that the
microcirculation of the renal medulla functions as a countercurrent
exchanger that traps NaCl and urea deposited to the interstitium by the
loops of Henle and collecting ducts, respectively. Early hypothetical descriptions of this process envisioned a system in which descending vasa recta (DVR) and ascending vasa recta (AVR) are parallel tubes that
equilibrate by diffusion. According to that notion, blood flowing from
the corticomedullary junction toward the papillary tip in DVR is
concentrated by diffusive influx of NaCl and urea, and, conversely,
blood flowing away from the papillary tip toward the
corticomedullary junction in AVR is diluted by diffusive efflux. That
theory predicts that solute is trapped due to recycling between DVR and
AVR and that net rate at which solute is removed from the medulla
is primarily dependent on the AVR-DVR concentration difference at the
corticomedullary junction (13, 39, 40, 91, 145).
Micropuncture studies demonstrated that AVR plasma is more
concentrated than adjacent DVR plasma, providing key evidence that vasa
recta are countercurrent exchangers (38, 113, 114). In contrast, fluid obtained from descending thin limbs of Henle is more
concentrated than that from ascending thin limbs, supporting the
conclusion that the loops of Henle function as the countercurrent multiplier responsible for generating corticomedullary osmotic gradients (38-41). Subsequent studies unraveled an
unexpected degree of complexity. DVR and AVR wall structures were found
to be distinct and to have characteristics that vary as function of
corticomedullary axis (39, 59, 91, 121). In vivo
measurements of plasma protein concentrations in DVR unexpectedly
revealed efflux of water from the DVR lumen to the papillary
interstitium, a finding that presented two paradoxes. First, the
physiological benefit derived from depositing water from DVR to
medullary interstitium was enigmatic. Second, measurements of Starling
forces (hydraulic and oncotic pressure) failed to predict the observed
direction of DVR transmural water movement (113, 114).
Recent physiological investigations continued to
show unexpected complexity while shedding some light on the paradoxes
associated with DVR equilibration. Specifically, physiological and
immunochemical measurements verified that aquaporin-1 (AQP1) water
channels and the facilitated urea carrier (UTB) are significant
transport pathways in DVR endothelia. Unusual intracellular signaling
pathways have been found in DVR endothelia (94, 99, 106).
A few recent measurements of AVR properties have been obtained that
show striking differences from DVR (81-83, 90, 95-98,
133). Overall, AVR remain poorly characterized because they
cannot be isolated for in vitro studies. Taken together, these studies
of microanatomy, tubular-vascular relationships, and transport
properties demonstrate complexity and lead to the conclusion that the
depiction of vasa recta as simple diffusive "U-tube" exchangers
leads to conceptual errors. In this review, we will summarize the
pertinent literature and, to the extent possible, give functional
perspective to these observations.
In view of his numerous contributions to the urinary concentrating
mechanism, among which were the introduction of the countercurrent multiplier theory of Werner Kuhn and the companion countercurrent exchange theory of the medullary circulation to American readers and
the experiments by him and his coworkers testing those theories, we
wish to dedicate this review to Robert W. Berliner, who died February
6, 2002.
| |
TUBULAR-VASCULAR RELATIONSHIPS AND MICROANATOMY |
Outer medulla.
Blood flow to the renal medulla is largely derived from the
efferent arterioles of juxtamedullary glomeruli (9, 10, 59, 73, 91, 93, 99), in addition to which a portion of the flow may
traverse periglomerular "shunt" pathways (16) (Fig. 1). Afferent arterioles that supply
juxtamedullary glomeruli arise from the cortical interlobular arteries
at a steep recurrent angle. Those afferent arterioles are composed of
one to three layers of smooth muscle cells surrounding the media and
endothelial layers. Efferent arterioles of juxtamedullary glomeruli are
larger in wall thickness, diameter, and length than efferent arterioles of superficial glomeruli (9, 10, 39, 91). As efferent arterioles penetrate across the corticomedullary junction to the outer
stripe of the outer medulla, the muscular layer decreases and is
replaced by smooth muscle remnants known as "pericytes" (Fig. 1).
In an arrangement compared with a "horse's tail," juxtamedullary efferent arterioles then give rise to as many as 30 DVR
(73). The diameter of rat DVR is generally one-half that
of the parent efferent arteriole (~12 µm ID), but some larger
vessels continue beyond the outer medulla to perfuse the deep inner
medulla (51). When efferent arterioles become DVR, smooth
muscle is replaced by pericytes and the medial layer interposed between
smooth muscle and endothelium disappears. As DVR continue into the
inner medulla, the pericytes become sparse but are present
(101). The transition from arteriole to capillary is most
gradual in the DVR that penetrate furthest into the inner medulla
(39, 91, 121).
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 1.
Anatomy of the medullary microcirculation.
A: within the renal cortex, interlobular arteries, derived
from the arcuate artery, ascend toward the cortical surface.
Juxtamedullary glomeruli arise at a recurrent angle from the
interlobular artery. The majority of blood flow to the medulla
traverses juxtamedullary efferent arterioles; however, a fraction may
also be derived from periglomerular shunt pathways. In the outer stripe
of the outer medulla, juxtamedullary efferent arterioles give rise to
descending vasa recta (DVR) that coalesce to form vascular bundles in
the inner stripe of the outer medulla. DVR on the periphery of vascular
bundles perfuse the interbundle capillary plexus that supplies nephrons
(thick ascending limb, collecting duct, long looped thin descending
limbs, not shown). DVR in the center of the bundles continue across the
inner-outer medullary junction to perfuse the inner medulla. Thin
descending limbs of short looped nephrons may also associate with the
vascular bundles in a manner that is species dependent (see
TUBULAR-VASCULAR RELATIONSHIPS AND MICROANATOMY, not
shown). In the inner medulla, vascular bundles disappear and vasa recta
become dispersed between thin loops of Henle, collecting ducts, and
ascending vasa recta (AVR). AVR that arise from the sparse capillary
plexus of inner medulla return to the cortex by passing through outer
medullary vascular bundles. DVR have a continuous endothelium
(left inset) and are surrounded by contractile
pericytes. Number of pericytes decreases with depth in the medulla. AVR
are highly fenestrated vessels (right inset). As
blood flows toward the papillary tip, NaCl and urea diffuse into DVR
and out of AVR. Transmural gradients of NaCl and urea drive water
efflux across the DVR wall via aquaporin-1 (AQP1) water channels. The
increasing size of the circled "P" is to represent DVR plasma
protein concentration that rises with medullary depth. B:
microvessel morphology. Juxtamedullary efferent arterioles are large
muscular vessels that give rise to DVR. DVR in vascular bundles retain
contractile smooth muscle-like pericytes. In the deep medulla, DVR
pericytes become more sparse but the endothelium remains continuous.
DVR wall becomes fenestrated near the origin of the inner medullary
capillary plexus and retains this characteristic in AVR. [From
Urinary Concentrating Mechanism: Structure and Function by
Rex Jamison and Wilhelm Kriz, copyright 1982 by Oxford University
Press. Used by permission from Oxford University Press, Inc.
(39).]
|
|
The architectural arrangement of the outer medulla is characterized by
a striking division into vascular bundles and the interbundle region
(Figs. 1 and 2). The "simple" vascular bundle (rabbit, guinea pig,
dog, cat, monkey, human) is comprised of DVR and AVR in close
apposition with a minimum of surrounding interstitium (6,
51). AVR that lie within vascular bundles are largely those that
originate within the inner medulla. AVR formed within the outer medulla
from the capillary plexus of the interbundle region ascend directly to
the cortex without rejoining a vascular bundle. The capillary plexus of
the interbundle region arises from DVR that peel off from the periphery
of the vascular bundles as they pass through the inner stripe
(42, 43, 59, 73, 91). From this arrangement we infer that
countercurrent exchange in the outer medulla occurs between AVR
draining the inner medulla and DVR supplying both outer and inner
medulla. The "complex" vascular bundle (rat, mouse,
Meriones, Psammomys) differs from simple bundles
due to the variable incorporation of thin descending limbs of short
looped nephrons (6, 7, 52, 59, 91). In the rat, the thin
descending limb is situated on the periphery of the bundles, whereas in
the mouse and Psammomys, short looped thin descending limbs
are distributed throughout the vascular bundle (7, 42, 51,
53). In some species, complex vascular bundles coalesce in the
outer stripe to form "giant" bundles that traverse the inner stripe
(Fig. 2A). The latter is most
prominent in Psammomys and is to be contrasted with the
vascular bundle architecture in the outer medulla of other mammalian
species (Fig. 2, B-D) (42). Thus vascular
bundles that characterize the inner stripe provide for countercurrent
exchange between DVR and AVR returning from the inner medulla, and to a
variable degree that is species dependent between vasa recta and thin
descending limbs of short looped nephrons. In Psammomys, an
additional striking feature exists. The vascular bundle periphery is
brought into proximity with pelvic urine due to invaginations of the
pelvic epithelium. The possibility that vascular bundles exchange
solute with the pelvic urine must therefore also be considered.
View larger version (92K):
[in this window]
[in a new window]
|
Fig. 2.
Vascular bundles. A: photograph of the microvasculature
of the desert rodent Psammomys obesus obtained by arterial
injection of the kidney with Microfil. The distinct pattern of the
cortex, outer, and inner medulla is apparent. In this species, the
separation of the outer medulla into vascular bundles and the dense
capillary plexus of the interbundle region (*) is striking because vasa
recta coalesce into giant vascular bundles. OM, outer medulla; IM,
inner medulla. Designations on the original figure are c, cortex; TR,
transitional region (outer stripe of the OM); IS, inner stripe of the
OM; IZ, inner zone (IM). B and C: injection study
of vascular bundles in the OM of the rat. In contrast to
Psammomys, individual vascular bundles do not coalesce into
giant bundles. They are more evenly dispersed throughout the inner
stripe of the OM. This pattern is most typical of mammalian species
including the rat, mouse, and human. D: transmission
electron micrograph showing AVR and DVR in the vascular bundles of the
rat. AVR are fenestrated and DVR have a continuous endothelium. The
interstitial space separating the DVR and AVR is small. [From Bankir
et al. (7), Moffat and Fourman (73), Pallone
et al. (97).]
|
|
Inner medulla.
Shortly after passing the inner-outer medullary junction, the vascular
bundle architecture disappears and vasa recta become more evenly
dispersed among nephrons and collecting ducts (39, 73,
91). In the outer medulla, interstitial space between adjacent
vessels is minimal (Fig. 2B). From the inner-outer medullary junction to the papillary tip, the fraction of medullary tissue that is
interstitium increases from 5 to ~30% (49). Numerous inner medullary interstitial cells, arranged horizontally like rungs of
a ladder, are tethered between vessels and nephrons so that they might
inhibit axial diffusion and retard dissipation of corticomedullary
gradients (58). As in the outer medulla, AVR outnumber
DVR. Their ratio has been reported as 1.7 to 1 in hamsters and 2.3 to 1 in the rat (35, 67, 155). DVR terminate at various levels
in a sparse capillary plexus that coalesces to form AVR. DVR have a
continuous, nonfenestrated endothelium and zona occludens (72,
121). Toward the termination of the DVR, fenestrations appear
that characterize the wall of the subsequent capillary plexus and AVR.
In the inner medulla, the fraction of the AVR wall covered by
fenestrations is ~50%. That fraction decreases toward the outer
medulla to ~15-30% (39, 91). On electron micrographs, fenestrations have diameters of 530 to 1,000 Å and are
bridged by a 40-Å-thick diaphragm. The diaphragm has one or two
concentric rings interconnected by radiating fibers and a central
density (68).
| |
COUNTERCURRENT EXCHANGE GENERAL CONCEPTS AND EVOLUTION OF
UNDERSTANDING |
Countercurrent exchange in nature.
The idea of countercurrent exchange can be traced back to Claude
Bernard (14), who observed "wherever a peripheral artery flows alongside a vein there is likely to be a heat gradient between them and a transfer of heat from artery to
vein ... [This] ... thermal short circuit ... carries
some of the arterial heat back into the body before it reaches the
periphery" [cited by Scholander and Krog
(119)]. As pointed out by Scholander and Krog
(119), this simple arteriovenous arrangement can become
much more complex with the artery dividing into arterioles or
capillaries and the vein dividing into venules or capillaries to
provide an enormous area for heat exchange between arterial and venous
blood, the rete mirabile (Fig. 3). The
sloth is a slow-moving arboreal animal that inhabits treetops in
tropical jungles. Scholander and Krog (119) dissected the
brachial rete of a sloth, a 1-cm-thick bundle of parallel arteries and
veins with 20-30 arteries and fewer veins. A 4-cm portion of a
rete was freed, and a thermocouple probe was threaded along its length.
The temperature decline in a direction away from the body was 1°C per
centimeter of length, 30 times steeper than that along the human
brachial artery. When the venous blood flow returning through the rete
was slowed by constricting a ligature, the gradient was greatly
reduced.
View larger version (227K):
[in this window]
[in a new window]
|
Fig. 3.
Rete mirabile (the wonderful net). Section through the rete mirable
that provides countercurrent trapping of oxygen in the swim bladder of
the deep sea eel. There is an enormous surface area for countercurrent
exchange that completely prevents loss of oxygen from venous blood by
enabling diffusion into blood entering from arterioles. Micrograph
originally generated by C. Lloyd Claff in collaboration with P. F. Scholander. [From Scholander (118).]
|
|
Cold extremities constitute an important adaptation to heat economy in
many Artic mammals, such as the whale with its greatly expanded and
well-vascularized fluke and fins, and the artic wading bird, which has
a rete at the junction of its body and stiltlike legs. For the sloth,
the ambient temperature at the treetop level can fall 15°C at night
(118, 119). Given the sloth's slowness of movement, heat
conservation is at a premium (118). Besides conservation
of heat, countercurrent exchange has a wide variety of applications in
biology (Table 1). Countercurrent flow
between blood and seawater in fish gills maximizes extraction of
oxygen. The most dramatic example is found in the swim bladder of deep sea fish to regulate their buoyancy at depths of thousands of meters
(Fig. 3). The pressure of gas (primarily oxygen) inside the bladder
lumen is enormous, 
200 atmospheres, equaling the weight of the
surrounding water. Located in the bladder neck, the rete traps oxygen
to prevent its escape from the bladder. The "... outgoing veins,
highly charged with oxygen, give it up to adjacent incoming arteries"
(119). For example, the rete from a common eel weighs
~65 mg and has 100,000 arterial capillaries and about the same number
of venous capillaries. The capillaries are ~4-mm long, meaning the
cumulative total length of each kind of capillary is 400 m.
Because the capillary diameter is 7-10 µm, the total wall
endothelial surface area exceeds 20 cm2 (54).
Scholander (118) calculated that the oxygen pressure across a
rete 1 cm long is reduced by a factor of 3,000, which means that oxygen
would be completely extracted from the venous blood flowing out through
the bladder neck and returned to the inflowing arterial blood.
Countercurrent exchange in the renal medulla.
The idea that the counterflow arrangement of vasa recta enables
efficient exchange of solutes and water originated with Kuhn and his colleagues (31, 55) and Wirz and colleagues
(145-147). But for many, the introduction to
countercurrent multiplication and countercurrent exchange was, at least
in the United States, provided by Berliner et al. (13).
The countercurrent multiplier explained the exponential rise in osmotic
pressure of the renal medullary tissue from the corticomedullary
junction to the papillary tip. But this implied that the osmotic
pressure of blood entering the medulla also rises concomitantly, which
posed the dilemma that even if only 5% of renal blood flow entered the
medulla, the concentrating mechanism would have to concentrate a very
large volume of blood to concentrate a much smaller volume of urine, severely limiting the mechanism's efficiency. The explanation that the
medullary circulation functions as a countercurrent exchanger resolved
the dilemma. Wirz (146) wrote, "all the blood irrigating the medulla enters and leaves the medulla at the corticomedullary boundary, i.e., in an essentially isotonic [to systemic blood] region. In between it may adapt itself to the osmotic pressure of the
surroundings by a passive uptake (on its way down) and release (on its
way up) of osmotically active solutes." Berliner et al.
(13) described countercurrent exchange using the analogy of a heat exchanger, depicted in Fig. 4.
In Fig. 4A, water flowing at 10 ml/min passes a heat source
that supplies heat at a rate of 100 calorie/min. Accordingly, the
temperature of the stream increases from 30 to 40°C. Figure
4B shows a hypothetical idealized counterflow heat exchanger
created by opposing the limbs upstream and downstream from the heat
source. In that case, consistent with the requirement of energy
conservation, water exits the exchanger at a temperature 10°C higher
than the inflow, but an axial temperature gradient is created by the
warming of water before its arrival at the heat source by the
outflowing heated water. This large axial temperature gradient is
established and maintained by thermal diffusion between the two limbs
and by heat trapping due to countercurrent flow. The graph
inset compares the temperature along flow tubes in each
system. In Fig. 4C, countercurrent flow principle is applied to an idealized capillary loop in the medulla representing a descending vas rectum and ascending vas rectum. Note that the exchange occurs between capillary and interstitium of the medulla rather than between
two adjacent capillary loops. Solute diffuses from the medullary
interstitium to the blood flowing down the descending vas rectum. As
blood returns in the ascending vas rectum, the solute concentration
difference is reversed and solute diffuses into the interstitium. This
effectively "traps" solute in the interstitium by recycling between
ascending and descending capillary. In this way, blood circulates
through the renal medulla without "washing out" its hypertonicity.
The greater the solute permeability of the capillary, the more complete
the exchange. In the absence of a high solute permeability, at any
given level, osmotic equilibrium may not be completely achieved;
consequently, the contents of the blood at the cortical medullary
junction may be slightly hypertonic to the isotonic interstitium.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 4.
Principle of countercurrent exchange. A:
delivery of 100 calorie/min of heat into a stream of water flowing at
10 ml/min. The temperature rises from 30 to 40°C. B: in
the presence of trapping afforded by countercurrent diffusion of heat
from ascending to descending limb by a U-shaped configuration, a much
higher temperature is achieved than when the flow is through a straight
pipe (A). C: schematic of a medullary capillary.
Countercurrent flow in a capillary loop where solute (concentration in
mosmol/kgH2O) exchanges between the ascending and
descending limb via a common interstitium. Solute is recycled by
diffusing out of the ascending limb and into the descending limb.
[From Berliner et al. (13).]
|
|
This model was consistent with the experimental findings of Wirz
(145), Ullrich and Pehling (137), and
Gottschalk and Mylle (30) that all structures in the
rodent renal papilla are equally hypertonic. Later Jamison et al.
(38), using a preparation (109) that exposed
a greater portion of the rat renal papilla, found evidence that blood
flowing through DVR lags in attaining osmotic equilibrium with the
hypertonic surroundings.
Water uptake by the medullary microcirculation.
The foregoing models neglect the exchange of water in the medulla. In
reality, water is supplied to the medullary interstitium by
reabsorption from the pars recta of the proximal tubule in the outer
medulla, the descending thin limb of Henle's loop in the inner
medulla, and the collecting duct. Since the epithelium of the ascending
thin and thick limb of Henle's loop is impermeable to water and there
are no lymphatics in the inner medulla and few in the outer medulla,
any water added to the interstitium must be removed by AVR.
Ullrich et al. (136) derived a model for the
countercurrent exchanger similar to Berliner's except that solute is
added along the length of the exchanger instead of being added only at
the tip of the loop. The differential equations are the same
(126). In rodents, measurements of hydrostatic pressure in
the DVR averaged 17 mmHg (136, 137). Ullrich postulated
that this would drive water from DVR, short-circuiting water along with
permeable solutes to AVR and concentrating plasma protein in the DVR.
In accord with this hypothesis, the ratio of vasa recta plasma to
systemic plasma protein was found to be >1, ranging from 1.08 to 1.40. (However, as the authors noted, an alternative explanation for the high
plasma protein is ultrafiltration upstream in juxtamedullary glomeruli.) The authors predicted that addition of water in the AVR
would reduce plasma protein to its preglomerular capillary value.
Micropuncture of the vasa recta in antidiuretic rats (41, 113,
114) and hamsters (30) near the papillary tip
confirmed the findings of Ullrich et al. (136) that the
plasma protein concentration at the end of the DVR is elevated above
that of the blood entering the medulla. Sanjana and colleagues
(113, 114) analyzed the driving forces and transmembrane
volume movement. In young rats with a systemic plasma protein
concentration of 4.1 g/dl, micropuncture of DVR and AVR at the base of
the exposed papilla revealed mean plasma protein concentrations of 7.1 and 5.6 g/dl, respectively. Assuming the vasa recta are impermeable to
protein, this finding implies dilution of plasma proteins by fluid
uptake between the DVR and AVR, a finding corroborated by Zimmerhackl
et al. (155). In contrast, micropuncture of the DVR at the
base and tip of the exposed papilla revealed protein concentrations of
5.6 and 6.4 g/dl, respectively, indicating net fluid loss from the DVR.
Water uptake in the AVR exceeded water removal from the DVR. The
difference accounted for the water added to the medulla from the
descending limb of Henle and collecting duct and confirmed mass balance
for fluid volume movement in the inner medulla.
After the blood begins to ascend in AVR toward the cortex, the solute
concentration in the plasma lags in equilibration with the continuously
decreasing axial concentration of interstitial solute. At some point
the direction of the transendothelial small solute concentration
difference will reverse and the luminal solute concentration will
exceed that in the interstitium and, if anything, will augment the
Starling forces, favoring water uptake by the capillary. The analysis
is confounded, however, by anatomic and structural differences between
AVR and DVR. First, AVR outnumber DVR between 2 and 2.5 to 1 (35,
155). The transit time of plasma flow through AVR is thereby
increased, which allows more time for equilibration of small solutes
between plasma and interstitium. Second, the endothelium of AVR is
fenestrated so that reflection coefficients for small solutes are
likely to be lower than in DVR, which has a continuous endothelium.
Comparisons of oncotic pressure exerted by plasma protein with
hydraulic pressure revealed that volume efflux from the DVR occurred
despite an oncotic pressure that exceeds hydraulic pressure. This was
in apparent conflict with conventional wisdom with regard to
transcapillary forces that determine water movement across the
capillary endothelium, as derived by Starling (124)
![J<SUB>v</SUB><IT>=</IT>L<SUB>p</SUB>[(P<SUB>c</SUB><IT>−</IT>P<SUB>i</SUB>)<IT>−</IT>(<IT>&pgr;</IT><SUB>c</SUB><IT>−&pgr;</IT><SUB>i</SUB>)]<IT>=</IT>L<SUB>p</SUB>[<IT>&Dgr;</IT>P<IT>−&Dgr;&pgr;</IT>]](/content/vol284/issue5/fulltext/R1153/img001.gif)
|
(1)
|
where Jv is the transmembrane volume flux
per unit of membrane surface area, Lp is the hydraulic
conductivity, P is the hydraulic pressure, 
is the oncotic pressure,
and the subscripts c and i refer to values in the capillary lumen and
interstitium, respectively. As illustrated in the model exchanger (Fig.
4C) and confirmed by Jamison et al. (38),
plasma concentrations of "small" (nonprotein) molecules such as
NaCl lag in osmotic equilibration with the surrounding interstitium,
creating a transendothelial difference in concentration. Sanjana et al.
(114) hypothesized that such a gradient of small solutes
might provide the additional osmotic driving force required for volume
efflux from the DVR. According to nonequilibrium thermodynamics, volume
flux across a membrane is defined by the following equations (44)
|
(2)
|
where 
P is the transmembrane hydraulic pressure difference,
i is the transmembrane osmotic pressure difference
due to the ith solute, and 
i is the
reflection coefficient of the membrane to the ith solute.
The equation states that volume flux occurs in response to a
transmembrane hydraulic pressure difference and the sum of the
transmembrane osmotic pressures exerted by all solutes that are
osmotically active across the membrane. Applying this to the DVR
|
(3)
|
where pr and ss are the reflection
coefficients of the capillary membrane to proteins and small solute,
respectively, and pr and ss are the
transmembrane osmotic pressure due to protein and small solutes,
respectively. From Van't Hoff's law
|
(4)
|
where R is the universal gas constant and T is absolute
temperature. Css is approximately twice the
transmembrane difference in sodium concentration, reflecting its
univalent anion partner. RT = 19.3 mmHg/mM at T = 37°C. For
example, the ss = 0.07 for NaCl in capillaries of
the frog mesentery (21). It has been found that the NaCl
concentration of the blood in DVR in the inner medulla is less than
that of the surrounding interstitium due to a lag in equilibration (see
below). Assume the transcapillary membrane concentration difference in
NaCl is 25 mmol/l and ss = 0.07 for NaCl. This
would result in 67.5-mmHg driving force for volume efflux from the DVR
into the interstitium. To test this hypothesis, Pallone et al.
(98) administered diuretics to rats to eliminate the axial
corticomedullary osmotic gradient (and therefore eradicate the
transendothelial NaCl gradient between interstitium and DVR blood). As
predicted, volume efflux from the DVR was abolished. Accessible DVR
near the surface of the papilla were perfused with buffers differing in
osmolality from the interstitium. Perfusion with solutions made
hyperosmotic or hyposmotic to the interstitium, by addition or removal
of NaCl, was accompanied by water uptake into the capillary and efflux from the capillary, respectively (85).
Equation 3 describes transport of water across the DVR
wall as a whole, simulating it as though it occurs through a single pathway, the hydraulic conductivity of which is Lp. More
recently the discovery of water channel proteins, the aquaporins, has
provided the long sought after biophysical explanation for selective
water permeability of biological membranes (2, 3, 76-78,
103, 108, 115). It is now understood that AQP1 is expressed by
DVR endothelia and is the transport pathway across which small
hydrophilic solutes such as NaCl and urea drive water flux. As will be
discussed in subsequent sections, transport of water across the DVR
wall is more rigorously described by simulating parallel pathways. One
pathway is the highly selective AQP1 molecule (ss = 1.0) and a parallel pathway that conducts both water movement as
well as convective and diffusive flux of small solutes
(ss 
0) (77, 87, 89, 95, 135).
Expression of AQP1 in DVR has been hypothesized to play an
important role in the optimization of renal medullary countercurrent
exchanger function (87).
The finding that urea transport across the collecting duct in the
presence of AVP is much greater than can be explained by diffusion and
is reduced by phloretin and urea analogs led to the discovery of a
transporter that facilitates urea movement. In the last decade, two
families of urea transporters have been identified (5,
17), UTA and UTB. UTA isoforms are present in the collecting
duct and descending limb of Henle's loop (133), and UTB
is found in erythrocytes and DVR (8, 80, 97). In subsequent sections, the role of UTB (Fig.
5) and AQP1 (Fig.
6) in the optimization of urinary
concentrating ability will also be considered.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 5.
Vasa recta solute permeabilities. A:
[14C]urea permeability (PU, ordinate) vs.
22Na permeability (PNa, abscissa) is shown for
outer medullary DVR (OMDVR) isolated from Sprague-Dawley rats and
perfused in vitro and inner medullary DVR and AVR (IMDVR, IMAVR)
perfused on the surface of the exposed papilla of Munich-Wistar rats in
vivo. Dashed line represents identity. Note that PU and
PNa are highly correlated and nearly equal in the IM but
show no correlation in OMDVR. Also note that the PNa of
some OMDVR is low but that PU is uniformly very high.
Dissociation of PNa and PU in OMDVR results (at
least in part) from expression of the UTB facilitated urea carrier.
B: PU measured in isolated OMDVR at baseline, or
in the presence of thiourea, methylurea (50 mM each, bath and lumen),
or phloretin (0.5 mM) and in a recovery period after washout of the
inhibitor. In each case, PU is reduced.
* P < 0.05 vs. baseline. See
Countercurrent exchange in the renal medulla for further
discussion. [From Pallone et al. (97).]
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 6.
Osmotic water permeability (Pf) of OMDVR.
A: Pf was measured in glutaraldehyde fixed rat
OMDVR by measuring transmural water flux generated by imposing a
bath-to-lumen gradient of NaCl. Sequential measurements in controls
revealed stability, whereas exposure to
p-chloromercuribenzene sulfonate (pCMBS, 2 mM) reduced
Pf to nearly zero. Glutaraldehyde fixation was necessary to
prevent deterioration of the vessel caused by either the large osmotic
gradient or prolonged pCMBS exposure. B: Pf was
measured in AQP1 null (/) or replete (+/+) murine OMDVR by imposing
bath-to-lumen gradients of NaCl, urea, glucose, or raffinose. Deletion
of AQP1 reduced DVR Pf, measured by driving water flux with
NaCl, from a control value of ~1,100 µm/s to nearly zero. Water
flux driven by raffinose (molecular wt 564) was markedly reduced
in the AQP1(/) vessels, compared with AQP1(+/+) vessels, but
remained unexpectedly high in the former. Similarly, glucose (molecular
wt 180) and urea (molecular wt 60) gradients drove measurable water
flux across AQP1(/) DVR. See Water uptake by the medullary
microcirculation for further discussion. [From Pallone et al.
(87).]
|
|
| |
TRANSPORT PROPERTIESGENERAL DEFINITIONS |
To understand the physiology of microvascular exchange in the
renal medulla and review the associated literature, one must grapple
with a few of the fundamentals of membrane transport theory. As
described above in association with Equations 1-4, DVR
equilibrate with the medullary interstitium by passive transport of
solutes and water through a variety of pathways. In this section, we
provide definitions of key parameters that define the properties of
those pathways. Measurement of those parameters has been the goal of many studies (Tables 1 and 2). Some description of the essentials is
provided in the APPENDIX and the reader is directed to
authoritative sources (4, 19, 26, 44, 69, 71, 102, 120,
144).
Water and solutes permeate the walls of microvessels, including DVR, by
passive convection and diffusion, driven by gradients of potential
energy provided by transmural differences in hydrostatic and
osmotic pressure (19, 69, 71). Quantitative analysis of
passive transport is based on nonequilibrium thermodynamics, which
states that fluxes through a membrane are proportional to driving
forces if they are small enough and if the system is not too far from
equilibrium (19, 44, 120). Permeability coefficients quantify relationships between transmembrane fluxes and forces or
between different fluxes. Hydraulic permeability (Lp)
relates the total flux of solvent (water) plus solute through a
membrane (volume flux, Jv) to the difference in
hydrostatic pressure between the two sides of the membrane (P).
Lp equals (Jv/P), when the transmembrane difference in solute concentration (C) and therefore in osmotic pressure difference () is 0. The resistance of a membranous pathway to transport of water is alternately expressed as
osmotic water permeability (Pf). Lp and
Pf are related; Pf = (Lp
Vw)/(RT), where Vw is the partial molar volume
of water.
Diffusional permeability to a solute (Ps) relates the net
molar flux of solute through a membrane (Js) to
the transmural concentration difference, C, when transmural volume
flux (Jv) and therefore convective solvent drag
is zero. Under these conditions, Ps simply equals
(Js/C) and can be viewed as the
"resistance" of the membrane to diffusion of the solute. Transport
of solute across a membrane can have both diffusional and convective
components the directions of which need not be the same. Equations that
describe this more complex scenario are provided in the
APPENDIX.
Osmotic reflection coefficient (d) is a property of a
membranous pathway that describes the selectivity of the pathway for solvent vs. solute. d Can take on values between zero
and one. d Is one for a semipermeable membrane that
sieves or "reflects" all solute from solution, but zero for a
nonselective membrane that does not distinguish between solute and
solvent. An ultrafiltration coefficient (f) is the ratio
of the convective solute flux reflected at a membrane to that carried
through the membrane, given by [1 (Js/JvC1)],
where C1 is the solute concentration at the upstream surface of the membrane and C is zero. For practical purposes in
physiological dilute solutions, d and f
are equal. The equality is, however, only approximate for nonideal
solutes, such as albumin (19, 69). As illustrated by
Equations 2 and 3, when the reflection coefficient to the ith solute (i) < 1.0, a transmural gradient of the solute will exert less than its total
ideal osmotic driving force for water movement. When
i = 1, solvent traversing the membrane will be
rendered solute free at the downstream membrane surface (complete
sieving). Conversely, when i = 0, movement of water
across the membrane carries solute freely, without restriction.
| |
TRANSPORT OF WATER THROUGH THE DVR WALL |
Water moves through the walls of DVR via pathways of at least two
kinds (77, 89, 90, 135). Analysis of the permeabilities of
DVR indicates that a "shared" transmural pathway for water and
hydrophilic solutes exists in parallel with a "water only" pathway
( 1.0) that excludes hydrophilic solutes.
Shared pathway.
Evidence for a shared pathway conducting diffusion of hydrophilic
solutes through the walls of DVR comes from measurements of the
correlations between the diffusional permeabilities of these
microvessels to 22Na (PNa) and to tritiated
water (PD), 36Cl, [3H]raffinose
(Praf), [14C]urea, and
[14C]inulin (90). The simplest
interpretation of correlated variations in diffusional permeability to
hydrophilic solutes is that they arise from variations in a shared
aqueous (porous) pathway. Simultaneous measurement of permeability to
two solutes was obtained by perfusing DVR in vitro with pairs of
radioactive tracers and calculating both permeabilities (e.g.,
PNa and Praf) from lumen-to-bath efflux using
dual isotope detection methods in the perfusate and collectate (Figs.
5-8, Table
2).
From Fick's first law (see APPENDIX), the diffusional permeability of a porous membrane to a solute (Ps) is given
by
|
(5)
|
where Dp represents the diffusion coefficient of the
solute inside the membrane pores, the partition coefficient (
) is
the equilibrium ratio of solute concentration in a pore to that in bulk
solution, and Ap is the fraction of the membrane area
occupied by pores and x is pore length (19, 66). Using
Equation 5, it can be predicted that, for diffusion of small
hydrophilic solutes in large pores, the ratio of the permeabilities to
those solutes should be equal to the ratio of their diffusion
coefficients in bulk solution. This was verified in DVR for several
pairs of tracers (Table 2) (90). It was experimentally
possible to demonstrate the correlations in rat DVR because the
permeability of individual vessels varies. The variation in
Ps between DVR is probably attributable to variations in
Ap/x, because the shared pathway in these microvessels does not significantly sieve small hydrophilic solutes (see below).
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 7.
Perfusion rate dependence of permeability measurements.
[3H]raffinose permeability (A) and
[14C]inulin permeability (B) are shown as a
function of perfusion rate of in vitro isolated, perfused DVR from
outer medullary vascular bundles of the rat. Change in permeability is
rapidly reversible upon lowering the perfusion rate (data not shown).
Heavy line connects mean ± SE of individual points.
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 8.
Correlations of solute permeability in OMDVR.
A: dual isotope perfusions of DVR were performed with
22Na and either 36Cl or
[3H]raffinose to measure simultaneous permeability.
Permeability to 36Cl or [3H]raffinose
(ordinate) is highly correlated with that to 22Na
(abscissa). Strong correlation and zero intercept are consistent with
permeation of these solutes via a shared pathway. B: dual
isotope perfusions of DVR were performed with
3H2O and 22Na. Permeability to
these isotopes is correlated but the intercept is nonzero. The latter
finding is consistent with the interpretation that
3H2O permeates the DVR wall through a pathway
shared with 22Na and through an additional independent
pathway, likely AQP1.
|
|
In contrast to AQP1-mediated water transport (Fig. 6), the shared
pathway also conducts most of the transmural convection (Jv) driven by oncotic pressure differences
across the walls of DVR (135). Evidence for this comes
from paired estimates of Jv and Praf
in isolated DVR perfused with a high molecular weight fluorescent
volume marker plus [3H]raffinose and intermittently
exposed to high concentrations of albumin (135). The
product of hydraulic conductivity and reflection coefficient to albumin
(Lpalb) was calculated from volume flux (Jv) driven by the known transmural osmotic
pressure difference provided by albumin (alb) when
P was negligible. Lpalb correlates with
Praf in DVR, with an intercept close to zero (Fig.
9), indicating that most
Jv driven by alb goes through
the shared pathway. The shared pathway is insensitive to mercurial
compounds, unlike the exclusive water pathways described below, because
p-chloromercuribenzenesulfonate (pCMBS; see Fig. 6) does not
change Lpalb in glutaraldehyde-fixed DVR
(89). Fixation of DVR by brief exposure to glutaraldehyde prevents damage by mercurials or hyperosmolar solutions and does not
change permeabilities (77, 87, 89).
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 9.
Correlation of OMDVR hydraulic conductivity
(Lp) and 3[H]raffinose permeability.
Lp was measured by driving water flux across in vitro
perfused OMDVR with hyperoncotic albumin. Diffusional permeability to
3[H]raffinose was measured by lumen-to-bath efflux of the
isotope. Product of Lp and osmotic reflection coefficient
to albumin correlates with Praf. This suggests that the
water flux occurs through a pathway shared with raffinose. See
Shared pathway for further discussion.
|
|
Flow appears to increase the hydraulic and diffusional permeabilities
of the shared pathway in DVR by a mechanism that is unknown in these
microvessels (Fig. 7). Lpalb,
Praf, and PNa correlate with perfusion rate in
isolated DVR, but the ultrafiltration coefficient for albumin does not
(89, 90, 135). Mean Lp values reported for DVR
are >1.4 × 10
6
cm · s1 · mmHg1
in vivo (96) and 1.56 × 106
cm · s1 · mmHg1
in vitro (if alb is 1) (135) (Table
3).
Molecular sieving by the shared pathway in DVR appears to be slight for
small hydrophilic solutes but has not been systematically investigated.
The shared pathway in rat DVR apparently offers little restriction to
diffusion or entry of hydrophilic solutes up to the size of inulin,
because diffusional permeability ratios for pairs of these solutes
resemble corresponding ratios of free diffusion coefficients (D)
(90). If solutes enter the shared pathway with little
restriction ( close to 1, Equation 5), then this implies
that their solutions undergo little ultrafiltration and exert only
small fractions of their total osmotic pressures across this pathway.
Mathematical modeling indicates that osmotic reflection and
ultrafiltration coefficients are ~(1 )2 for
porous membranes or for fibrous networks (such as the glycocalyx lining
microvessels) and therefore are small if is close to one (19,
29, 69, 71). Osmotic reflection coefficients for NaCl solutions
at the walls of rat DVR are indeed small, estimates being 0.032 in
vitro (135) and <0.05 in vivo (85). These
values are calculated from the relative abilities of NaCl and albumin solutions to drive transmural volume flux in unfixed DVR, assuming that
alb is one. They are probably overestimates of the
osmotic reflection coefficient of NaCl (s) at the shared
pathway, because they describe molecular sieving by whole DVR, which
occurs at highly selective exclusive water pathways as well as at the
shared pathway (see below and APPENDIX). In fact, transport
of water across the DVR wall may be better described by simulating
parallel transport through the shared pathway (s 0)
and water channels (s 1.0).
Molecular sieving by a shared pathway in DVR is poorly defined even in
microvessels from AQP1 knockout mice, which retain only minimal
exclusive water pathways (87). Osmosis drives volume efflux from these DVR through a mercurial-insensitive pathway, which
appears to show increased sieving of progressively larger hydrophilic
solutes. Hyperosmolar solutions of NaCl are ineffective, but urea,
glucose, and raffinose are increasingly able to drive volume efflux
from unfixed and fixed AQP1 knockout DVR (Fig. 6B). pCMBS
does not inhibit raffinose-driven volume efflux from these microvessels. Similarly, AQP1 knockout DVR apparently do not
ultrafilter luminal 22Na, but do retain some
[3H]raffinose and [14C]inulin during volume
efflux driven by hyperosmolar raffinose. Unfortunately, mathematical
simulations of ultrafiltration and transmural diffusion along mouse DVR
do not yield reliable estimates of osmotic reflection or
ultrafiltration coefficients for small hydrophilic solutes because of
the high diffusional permeabilities of mouse DVR to these tracers
(87). Hence, the appearance of size-dependent molecular
sieving by AQP1 knockout DVR, although attributable to a shared
pathway, is equally consistent with complete sieving at remaining
exclusive water pathways, combined with slower transmural diffusion of
larger solutes.
In contrast with small hydrophilic solutes, the macromolecule albumin
undergoes considerable sieving at the shared pathway in DVR. The mean
ultrafiltration coefficient of albumin solutions at whole DVR is 0.89 (and not significantly different from 1), according to retention of
fluorescently labeled perfusate albumin during volume efflux driven by
unlabeled albumin (135). This probably is close to the
ultrafiltration coefficient of albumin solutions at the shared pathway,
because this pathway dominates the Lp of DVR.
Exclusive water pathways.
Water apparently diffuses through the walls of DVR via pathways
that exclude hydrophilic solutes, as well as via shared pathways, because the permeability to tritiated water (PD) of these
microvessels is always high, even when PNa is low
(90) (Fig. 8B).
Exclusive water pathways also conduct most of the convection driven
through the walls of DVR by transmural osmotic pressure gradients due
to small hydrophilic solutes (85, 89). Evidence for this
is that mathematical simulations of ultrafiltration and transmural
diffusion along rat DVR, during volume efflux driven by hyperosmolar
NaCl, most accurately predict the observed retention of
22Na and [3H]raffinose when
ultrafiltration and osmotic reflection coefficients are assumed to
be one at the convective pathway (135) (Fig.
10).
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 10.
Estimation of reflection coefficients by measurement of
molecular sieving of 22Na (A) and
[3H]raffinose (B) by the DVR wall. OMDVR from
rats were perfused with 22Na and
[3H]raffinose and then subjected to a bath-to-lumen
concentration gradient of NaCl to drive efflux of water.
Collectate-to-perfusate concentration ratios of the isotopes
(RNa and Rraf) were measured (abscissa) and
predicted from a mathematical simulation (ordinate) assuming reflection
coefficients of either 0.1 or 1.0. A reflection coefficient of 1.0 (complete molecular sieving) was consistent with the data and the
behavior expected from AQP1-mediated transcellular water movement. See
Exclusive water pathways for further discussion.
|
|
The walls of DVR therefore seem to contain exclusive water pathways in
parallel with a shared pathway that sieves small hydrophilic solutes
poorly and macromolecules well. For parallel pathways (19), the volume flux driven through the walls of DVR
(Jv) by transmural differences in
concentration of small hydrophilic solutes (css) can be
described by
|
(6)
|
where Vw is the partial molar volume of water,
ss is the osmotic reflection coefficient for a small
hydrophilic solute, A is the fractional area of a pathway, and the
subscripts a, w, and p denote values for the whole microvessel for
exclusive water and for shared pathways, respectively, so that
|
(7)
|
Mean osmotic permeabilities of exclusive water pathways
(Pf,wAw) in unfixed and fixed DVR from rats are
between 900 and 1,300 µm/s (89, 135). These estimates of
Pf,wAw ("apparent Pf") of DVR
are calculated from Jv driven by
Css due to NaCl, because of the evidence that osmotic
reflection coefficient of NaCl is one at the exclusive water pathway
(ss,w) but low at the shared pathway
(ss,p). This evidence implies that NaCl drives volume flux mainly through the exclusive water pathway, because it exerts little effective osmotic pressure across the shared pathway. From Equation 6, Jv Pf,wAwVCss if
ss,p 0. Flow apparently does not modulate
Pf,wAw in glutaraldehyde-fixed DVR
(89), as it does Pf,a in the unfixed
microvessels (135) or PNa in unfixed (90) or fixed (89) DVR.
The osmotic permeability of exclusive water pathways in DVR
(Pf,wAw) is one order of magnitude lower than
that of the shared pathway (Pf,pAp) (Table 3).
This follows because an osmotic permeability for whole DVR
(Pf,a) of 16,700 µm/s can be calculated from the mean
Lp of 1.56 × 106
cm · s1 · mmHg1
for microvessels in vitro (89, 135). These permeabilities are of the shared plus exclusive water pathways in DVR, because this
Lp is calculated from Jv driven by
albumin solutions, which apparently have osmotic reflection
coefficients close to one at both pathways.
The low osmotic reflection coefficient of NaCl solutions at whole DVR
(85, 135) is consistent with the osmotic permeabilities and molecular sieving properties attributed to these shared and exclusive water pathways. For parallel pathways (19), from
Equations 6 and 7
|
(8)
|
This predicts that ss,a 0.06, if
ss,p is zero, ss,w is one,
Pf,wAw is 1,000 µm/s, and Pf,a is
16,700 µm/s, which agrees reasonably well with experimental values
for ss,a < 0.05 for NaCl solutions (Table 3)
(85, 135). Note that ss,a (Equation 8) is the same as ss in Equation 3 as
originally applied to the DVR wall by Sanjana et al.
(114).
Identification of exclusive water pathways in DVR begins with the
observation that they are mercurial sensitive, unlike the shared
pathway. pCMBS strongly inhibits volume efflux driven by hyperosmolar
NaCl from fixed DVR from rats (89) (Fig. 6A)
and significantly reduces PD without changing
PNa (77). Similarly, pCMBS abolishes volume
efflux driven by NaCl from wild-type mouse DVR and reduces that driven
by raffinose (87). This suggests that aquaporins form a
transcellular exclusive water pathway in DVR, although mercurials do
not block all (2) or only (80) aquaporins.
AQP1 is highly selective for water, mercurial sensitive (2,
3), and expressed by DVR (77, 78) in sufficient
quantity to account for exclusive water pathways (89).
Polyclonal antibodies to AQP1 label the plasma membranes (including
caveolae) of the continuous endothelium of DVR in the inner medulla of
rat kidney, but not the surrounding pericytes or the fenestrated
endothelium of AVR (77). The AQP1 content of rat DVR,
measured by enzyme-linked immunosorbent assay, predicts an osmotic
permeability of 1,344 µm/s, if AQP1 is equally distributed between
luminal and abluminal endothelial plasma membranes in series
(89<
TOP
ABSTRACT
INTRODUCTION
GENERAL...
