Renal Cortical and Medullary Microcirculations: Structure and Function

Cortical Microcirculation

The basic vascular pattern of the kidney is preserved across mammalian species ( Figure 24.1 ). The renal artery branches into interlobar arteries that ascend within the renal pelvis to enter the parenchyma. In multipapillate organisms, the interlobar arteries travel toward the cortex along the columns of Bertin. These vessels change direction and follow an arc-like course near the corticomedullary border to become arcuate arteries. The arcuate artery gives rise to the interlobular arteries that ascend, in radial fashion, toward the cortical surface. Afferent arterioles arise from interlobular arteries at angles that vary with their cortical depth. Afferents that supply deep glomeruli near the corticomedullary junction (juxtamedullary glomeruli) leave the interlobular artery at a recurrent angle. In contrast, superficial afferent arterioles that supply glomeruli near the surface of the kidney line up with the interlobular artery at its termination.

In some species (rat, cat, dog, and Meriones ) a pair of intra-arterial “cushions” exist as parallel ridges that project from the origin of the juxtamedullary afferent arteriole into the lumen of the parent intralobular artery ( Figure 24.2a ). The cushions are 18–30 μm in length, composed of a ground substance in which smooth muscle cells are embedded and covered by a layer of continuous endothelium ( Figure 24.2b ). The cushions are ideally placed to regulate blood flow distribution within the cortex. Given that blood flow to the renal medulla arises largely from the efferent flow of juxtamedullary glomeruli, it is also plausible that the cushions play a role in the regulation of blood flow to the medulla. The cushions have been hypothesized to separate plasma and red blood cells, and alter medullary hematocrit by “skimming” plasma from a red cell-free layer near the vessel wall.

The structure of afferent and efferent arterioles varies with cortical location. Afferent arterioles are composed of one to three layers of smooth muscle cells. Muscle and elastic tissue diminish near the glomerulus and the media is replaced by granular cells of the juxtaglomerular apparatus. The diameters of efferent arterioles vary from as small as 12 μm in the superficial cortex to 28 μm for juxtamedullary glomeruli. Efferent arterioles in the superficial cortex branch to form a dense peritubular capillary plexus. Ten different branching patterns have been described. In contrast to the efferent arterioles of superficial glomeruli, those of juxtamedullary glomeruli most frequently cross the corticomedullary junction to enter the outer stripe of the outer medulla, where they give rise to descending vasa recta (DVR). DVR are transitional vessels along which smooth muscle is replaced by contractile pericytes. A small fraction of blood flow to the renal medulla may bypass juxtamedullary glomeruli in “shunt” vessels (dashed line, Figure 24.1 ). Casellas and Mimran have described variations of those shunts as: (1) branches of afferent arterioles; (2) continuous vessels from which afferent arterioles arise as side branches; (3) short vascular connections between afferent and efferent arterioles; and (4) pelvic arterioles derived from afferent arterioles near the renal hilus ( Figure 24.3 ). Several reviews of arteriolar patterns in the renal cortex have been written.

The bulk of glomerular filtrate is reabsorbed by the proximal convoluted tubule. That reabsorbate is conducted through the cortical interstitium and taken up by peritubular capillaries. The peritubular interstitium of the cortex has been divided into “narrow” and “wide” portions comprising 0.6 and 3.4% of tissue volume. The narrow interstitium is flanked by a highly fenestrated capillary wall on one side and the basolateral membrane of the PCT on the other ( Figure 24.4 ). Since only 26% of the tubular surface faces the narrow interstitium it follows that substantial quantities of fluid must flow from the wide to the narrow portion, implying that hydrostatic pressure gradients exist within the cortical interstitium.

Medullary Microcirculation

Blood flow to the renal medulla is supplied by DVR. Descending vasa recta (DVR) arise largely from efferent arterioles of juxtamedullary glomeruli and supply all blood flow to the renal medulla ( Figure 24.1 ). The afferent arterioles that supply juxtamedullary glomeruli arise from interlobular arterioles (cortical penetrating arterioles) at recurrent angles. Muscular intra-arterial cushions exist at the origins of afferent arterioles ( Figure 24.2 ), particularly near the corticomedullary junction, and have been proposed to participate in reduction of medullary hematocrit by “plasma skimming” (see below). Juxtamedullary efferent arterioles are larger, longer, and have a more robust smooth muscle layer than those derived from superficial or midcortical glomeruli ( Figure 24.5 ). They often divide into branches that either remain in the cortex to supply cortical peritubular capillaries or, alternatively, descend into the medulla to form DVR. DVR are about one half of the diameter of parent juxtamedullary efferent arterioles, in the range of 12 to 18 μm; some may be as large as 20 μm. DVR branch from their parent efferent arteriole in the outer stripe of the outer medulla and then coalesce within vascular bundles in the inner stripe of the outer medulla ( Figure 24.1 ). Larger diameter DVR lie in the center of vascular bundles and penetrate to the deepest regions of the inner medulla. DVR are transitional vessels, wherein smooth muscle is replaced by contractile pericytes. Pericytes become increasingly scarce with medullary depth, but are retained well into the inner medulla ( Figure 24.6 ). DVR have a continuous endothelium with tight junctions. In contrast, ascending vasa recta (AVR) that arise from DVR are highly fenestrated ( Figure 24.7 ). DVR peel off from the periphery of the vascular bundles as they pass through the inner stripe to supply the interbundle capillary plexus ( Figure 24.8a,b,c ). AVR within vascular bundles are only those that originate from the inner medulla. Blood flow from the outer medullary interbundle capillary plexus returns to the cortex without rejoining vascular bundles. Thus, countercurrent exchange in the vascular bundles of the inner stripe of the outer medulla involves all DVR, but only those AVR that return from the inner medulla.

The most striking characteristic of the outer medullary circulation is its separation into vascular bundles and the dense capillary plexus that supplies the interbundle region of the inner stripe ( Figures 24.1, and 24.8 ). Species variation exists with respect to the association of nephrons with vascular bundles. The “simple” vascular bundle of the rabbit, guinea pig, dog, cat, monkey, and man is comprised only of DVR and AVR, excluding nephrons. The “complex” vascular bundle of some rodents incorporates the descending thin limbs of short looped nephrons (nephrons that return from the inner–outer medullary junction). The degree to which thin descending limbs of Henle are incorporated into vascular bundles varies with species, and is highly developed in the mouse ( Figure 24.9 ). Psammomys obesus is a desert-dwelling rodent in which vascular bundles combine in the outer stripe of the outer medulla to form “giant” vascular bundles ( Figure 24.8a ). By anatomical inference, the parallel arrangement of DVR within vascular bundles of all species contributes to the regulation of regional blood flow distribution in the kidney. Constriction of DVR on the vascular bundle periphery should favor perfusion of the inner medulla. Conversely, constriction of DVR in the bundle center should preferentially favor flow to the capillary plexus of the interbundle region. Experimental evidence that supports such differential regulation of perfusion of the outer and inner medulla is sparse, because few studies have simultaneously measured outer and inner medullary blood flow.

The vascular bundles that are characteristic of the inner stripe of the outer medulla disappear below the inner–outer medullary junction. Throughout the medulla, AVR are larger and more numerous than DVR. As a consequence, during passage of blood from the juxtamedullary efferent arteriole to DVR and then AVR, single vessel flow rate falls as overall microvessel cross-sectional area increases. The latter increases transit time, presumably to favor greater equilibration of solute concentrations between AVR blood and interstitium. Outer medullary vascular bundles have little interstitial space. In contrast, the fraction of medullary cross-section attributable to interstitium rises substantially toward the deepest regions of the inner medulla, particularly near the papillary tip. In some species, renal medullary interstitial cells (RMIC) of the inner medulla appear to be tethered between thin limbs of Henle and vasa recta ( Figure 24.10 ). It is likely that the horizontal arrangement of RMIC helps to preserve corticomedullary solute gradients by limiting axial diffusion along the medulla. RMIC have receptors for vasoactive peptides such as angiotensin II, bradykinin, and endothelin. In addition, they release vasoactive agents such as PGE ₂ and medullipin. RMIC are contractile and respond to various paracrine factors.

Three-dimensional computer reconstructions of images derived from immunostained serial sections have yielded insight into relationships between tubules, collecting ducts, and the vasculature in rats and mice. Specifically, collecting ducts (CD), descending thin limbs of Henle (DLH), ascending thin limbs of Henle (ALH), descending vasa recta (DVR), and ascending vasa recta (AVR) are identified by their respective expression of aquaporin 2 (AQP2), aquaporin 1 (AQP1), chloride channel (ClCK1), urea transporter type B (UTB), and the fenestral protein, PV-1. Thin DLH frequently lack AQP1 expression, and are identified as thin limbs lacking ClCK1 but expressing α,β-crystallin, an antigen that is present along the entire loop of Henle. Thin DLH include an entirely AQP1-null group that turns early in the inner medulla, and longer limbs that are AQP1-positive for the first ~40% of their length; the latter turn to form thin ALH within deeper regions of the inner medulla. More than 50% of the AQP1-positive DLH also express ClC-K1. AQP1-positive segments have semilunar cell bodies that regularly jut into the lumen. Murine tubulovascular relationships in the outer medulla may exhibit an important characteristic, in that some thick ascending limbs of long looped nephrons may lie within vascular bundles. In that species, short looped thin DLH are known to be incorporated into the more peripheral parts of vascular bundles.

Pannabecker and colleagues have described clustering of collecting ducts that eventually coalesce to form single large collecting ducts in the deep medulla ( Figure 24.11 ). The tubules and vessels that surround the collecting duct clusters occupy structured patterns. DVR and DLH occupy regions outside the clusters, while AVR and ALH are diffusely distributed both within the central regions of the CD clusters and throughout the surrounding inner medulla. Within clusters, four AVR abut individual CDs. AVR closely approach the CD wall and appear to be tethered to it. That arrangement may be critical to vascular reabsorption in the medulla, allowing interstitial pressure to exceed AVR luminal pressure without inducing collapse. The implications of those tubulovascular relationships have been examined in mathematical simulations. The organization of the outer medulla into vascular bundles and the peribundle region enhances the delivery of high osmolality fluid to the inner medulla by long looped DLH and CD. Reduction of the number of DVR that reach the inner–outer medullary junction is predicted to favor enhancement of urinary concentration, as is a high AVR solute permeability. The striking observation of thick ascending limbs of Henle within vascular bundles of the mouse may have important implications for urinary concentration, because sodium chloride reabsorption from those structures might raise vicinal osmolality to favor water uptake from AQP1-expressing DVR, thereby concentrating their contents en route to the inner medulla.

The Renal Cortex and Capillary Uptake of Tubular Reabsorbate

Cortical peritubular capillaries are fenestrated, have a large surface area and high hydraulic conductivity. It is generally accepted that fluid is driven into the cortical interstitium from the PCT due to the generation of a locally hypertonic fluid within the lateral intercellular space between PCT epithelial cells. The local hypertonicity results from the secretion of small hydrophilic solutes by proximal tubular cells. Dilution of the interstitium in the vicinity of the capillary wall with protein-free fluid both lowers interstitial oncotic pressure and raises interstitial hydraulic pressure. These effects generate Starling forces that favor capillary reabsorption.

A substantial body of evidence has demonstrated that modulation of cortical peritubular capillary oncotic pressure alters PCT reabsorption. Intra-aortic injection or peritubular perfusion of colloid free or hyperoncotic fluid leads to decreases or increases proximal reabsorption, respectively. While it seems inviting to surmise that protein oncotic pressure acts directly to enhance fluid movement out of the PCT, several lines of evidence suggest otherwise. Oncotic pressure changes fail to modulate reabsorption when active transport by PCT epithelia is inhibited, PCT reabsorption rates correlate with GFR (glomerulotubular balance) but not with interstitial Starling forces, hydropenia blunts the capacity of hyperoncotic albumin to enhance PCT reabsorption, and elevation of luminal pressure in the PCT fails to enhance reabsorption. The osmotic water permeability of the proximal convoluted tubule (P _f = 0.1 − 0.4 cm/s) is probably too low for small oncotic pressure changes to substantially affect transmembrane water flux. It has been suggested that peritubular oncotic pressure might modulate PCT volume reabsorption by affecting solute reabsorption rate or by enhancing paracellular backleak into the PCT lumen. More recently, attention has turned to mechanosensation of luminal shear forces as a transducer of glomerulotubular balance.

Whatever mechanisms converge to influence PCT reabsorption it is immense, and its removal from the interstitium is the task of the cortical microcirculation. Since lymphatics remove less than 1% of the reabsorbate, the route for return to the systemic circulation must be via peritubular capillaries. As reviewed by Aukland et al., the high oncotic pressure of postglomerular plasma cannot be invoked as the primary driving force for capillary uptake in all cases. Older rats in which renal lymphatic (and therefore presumably interstitial) protein concentration is equal to that of plasma continue to reabsorb tubular fluid. Furthermore, PCT reabsorption occurs even in kidneys perfused with colloid free solutions. Particularly in the latter case, it is apparent that interstitial pressure must exceed intracapillary luminal pressure to provide the driving force for transcapillary volume flux. It also follows that the peritubular capillaries must be tethered to the interstitium in a way that prevents an inwardly directed transmural pressure from collapsing the lumen.

The Renal Medulla and Countercurrent Exchange

Like the capillary bed of other organs, the renal medullary microcirculation supplies oxygen and nutrients to the surrounding tissue. Additionally, however, corticomedullary gradients of NaCl and urea must be preserved to enable urinary concentration. This task is accommodated by countercurrent exchange between DVR and AVR. Countercurrent exchange is an adaptation found throughout nature. The maintenance of high gas tensions in swim bladders of deep sea fish, and the minimization of heat loss from the extremities of aquatic and arctic animals relies upon this strategy.

The microcirculation of the renal medulla traps NaCl and urea deposited to the interstitium by the loops of Henle and collecting ducts. Countercurrent exchange provides the means by which blood flow through the medulla is concentrated and then diluted to preserve corticomedullary solute gradients established by countercurrent multiplication. The hypothesis that vasa recta are a purely “U-tube” diffusive countercurrent exchanger implies the following function. NaCl and urea diffuse from interstitium into DVR plasma en route from the corticomedullary junction toward the papillary tip. The solutes diffuse out from AVR plasma to be returned to the interstitium as blood returns to the cortex. That theory predicts that countercurrent exchanger efficiency will be enhanced if permeability to solute is high.

In fact, vasa recta probably do not function as a purely diffusive countercurrent exchanger. Several features point to greater complexity. Tubulovascular relationships in the outer and inner medulla differ markedly ( Figures 24.1, 24.9, 24.11 ), and the endothelium of DVR and AVR are continuous and fenestrated, respectively ( Figure 24.7 ). The discoveries of aquaporin-1 (AQP1) water channels and the facilitated urea carrier, UTB, in DVR endothelia (see below) shows that transcellular as well as paracellular pathways involving water and urea are involved in equilibration of DVR plasma with the interstitium. Efflux of water across the DVR wall to the medullary interstitium occurs across AQP1 water channels, and AQP1 excludes NaCl and urea implying that both water removal and diffusive influx of solute contribute to transmural equilibration. Expression of AQP1 within the vasculature has been found to vary with axial location, greater in those DVR that turn in the outer portion of the inner medulla.

Transport of Small Solutes and Water by Vasa Recta and Red Blood Cells

DVR Hydraulic Conductivity and Osmotic Water Permeability

The predominant pathway that conducts water efflux across the DVR is the AQP1 water channel. AQP1 but not other aquaporins are expressed by DVR endothelia. Diffusional water permeability (P _D ) of isolated, microperfused DVR, measured by efflux of ³ H ₂ O, was reduced by the AQP1 blocking mercurial agent p-chloromercuribenzene sulfonate (pCMBS). Dramatic confirmation was provided by the demonstration that osmotic water permeability (P _f ) of microperfused DVR, measured by driving water flux with transmural gradients of NaCl, was driven from ~1100 μm/s to nearly zero by pCMBS ( Figure 24.12a ). In contrast, when albumin rather than NaCl was used to drive water flux, P _f was much higher, ~16,700 μm/s and insensitive to pCMBS, implying that a different pathway conducts transmural volume flux driven by oncotic pressure. The results support the notion that NaCl and urea drive water flux across the DVR wall exclusively through AQP1 (contribution to total transmural water conductivity is P _f ~1100 μm/s), while hydraulic pressure and oncotic pressure drive most water flux through a high conductivity parallel pathway (paracellular or other). Mathematically, the AQP1 and parallel pathways are best stimulated as :

$J_{V\text{,}P}=L_{P\text{,}P}[\Delta P-{\sigma }_{pr}\Delta {\pi }_{pr}]{\text{and}\text{,}J}_{V\text{,}C}=L_{P\text{,}C}[\Delta P-\sum _i\Delta {\pi }_i]$

where the additional subscript “P” (probably pericellular) refers to the high conductivity pathway for which σ _NaCl = σ _urea = 0, and the subscript “C” refers to the transcellular AQP1 pathway for which σ _NaCl = σ _urea = 1. Hydraulic conductivity (Lp) and osmotic water permeability (P _f ) are related according to Lp = (P _f × V _w )/(RT), where V _W is the partial molar volume of water. Existing measurements of DVR osmotic water permeability are summarized in Table 24.2 . A rigorous discussion of the measurement of, and relationships between, these transport parameters has been provided.

Table 24.2

Hydraulic Conductivity, Osmotic Water Permeability and Solute Reflection Coefficients of Vasa Recta

Parameter	Driving Force	OMDVR	IMDVR	IMAVR	Reference
L p × 10 −613 (cm•s −780 •mmHg −780 )	Albumin Gradient		1.4 a
L p × 10 −613 (cm•s −780 •mmHg −780 )	Albumin Gradient	1.6
L p × 10 −613 (cm•s −780 •mmHg −780 )	NaCl Gradient	0.12 b
L p × 10 −613 (cm•s −780 •mmHg −780 )	Hydraulic Pressure			12.5

Parameter	Method	OMDVR	IMDVR	IMAVR	Reference
σ albumin	Sieving	0.89 c
σ albumin	Sieving			0.78
σ albumin	Osmotic			0.70
σ Na	Osmotic		<0.05 d	0.00 d
σ Na	Osmotic	~0.03 d
σ Na	Sieving	~1.0 e
σ Raffinose	Sieving	~1.0 e

a Assumes a reflection coefficient to albumin of 1.0.

b Evidence shows that transmural NaCl gradients drive water flux exclusively through water channels, whereas albumin drives water flux predominantly through water channels along with a small component via other pathway(s), see text and references.

c Not significantly different from 1.0.

d Measurement of σ _Na for the vessel wall as a whole.

e σ _Na , σ _Raffinose for the putative aquaporin-1 water channel pathway through which NaCl gradients drive water flux, see text and references .

The AQP1 knockout mouse provided additional confirmation of the role of AQP1 in DVR water transport. P _f of DVR in AQP1 knockout mice, driven by NaCl, was indeed very low ( Figure 24.12b ). An intriguing finding was that urea and larger solutes (raffinose, MW 594, glucose, MW ) drive significant water flux despite AQP1 deletion, a finding that implies the existence of a non-AQP1 route across which those solutes are osmotically active. A potential candidate for the non-AQP1, pCMBS insensitive pathway is the UTB urea transporter (see below) which is expressed by the DVR endothelium and can function as a water channel.

Insights from Modeling: AQP1 and the Enhancement of Exchanger Efficiency

Mathematical models of urinary concentration have played an important role in the evolution of our understanding. Simulation of both nephrons and the microcirculation is difficult, so that investigators prefer to account for one while neglecting the other. Vasa recta models typically assume specified corticomedullary solute concentrations, and simulate transport properties of the vessel wall. Wang and Michel revised this approach by specifying the rate of deposition of NaCl, urea, and water to the medullary interstitium as though they are generated within the interstitium. In agreement with electron probe measurements, they predicted an exponential increase in corticomedullary solute concentration in the medulla. A weakness of models that neglect simulation of loops of Henle and collecting ducts is that solute generation rates in the interstitium are assigned as inputs and interstitial solute concentrations calculated as predictions. Variations of blood flow and transport properties cannot affect the interstitial appearance of NaCl, urea, and water from nephrons as would occur in vivo . Convincing evidence has been provided that structure and properties of nephrons can abruptly vary with medullary depth.

Many key parameters needed to simulate microvascular exchange in the renal medulla (solute permeabilities, reflection coefficients, hydraulic conductivities) have been measured. That data has been combined with more complete simulations of transcellular pathways for urea and water transport to perform additional simulations. The models predict that AQP1 might play an important role to raise medullary interstitial osmolality by driving water efflux from DVR to the medullary interstitium across AQP1 water channels (see above); that water movement effectively shunts DVR plasma volume to the AVR reducing blood flow to the deep medulla; and that favors high diffusive exchanger efficiency in the deep inner medulla where urea is added to the interstitium from the collecting duct. Stated another way, it reduces the lag in equilibration that leads to solute “washout” from the deep medulla. The net effect is to enhance interstitial osmolality ( Figure 24.13 ). Interest in this intriguing prediction is heightened by the observation that transmural water flux can be driven across the wall of AQP1-null mice by solutes other than NaCl ( Figure 24.12b ). If the non-AQP1 pathway is important in vivo , it might also enhance shunting of water from DVR to AVR.

Facilitated Transport of Urea across DVR and RBCs

Transmural flux of urea across the DVR wall is complicated, because DVR endothelia express a facilitated urea carrier. Sodium and urea have similar free water diffusivity, and are therefore expected to have the same transvessel permeability if they diffuse, sterically unrestricted, through a large pore. In contrast to this, some outer medullary DVR have low or moderate P _Na but high P _U ( Figure 24.14 ). DVR permeability to ¹⁴ C urea can be partially inhibited by phloretin, pCMBS, and structural analogs of urea, verifying the presence of an endothelial carrier. Histochemical evidence and in situ hybridization studies have shown that the DVR urea carrier is the same as that in the RBC (urea transporter type B, UTB), and is distinct from the urea carrier in the thin limbs of Henle (UTA2) and collecting duct (UTA1, UTA3, UTA4).

Rat UTB carries the Kidd blood group antigen, has 62% identity to UTA2, and is expressed in RBCs, DVR endothelium, papillary surface, and pelvic epithelium of the kidney. The presence of UTB in the DVR endothelium and RBCs facilitates medullary urea recycling. Urea tends to exit the renal medulla in AVR plasma and RBCs. To prevent associated dissipation of corticomedullary urea gradients, urea recycles from AVR into DVR plasma and RBCs via UTB, and into thin limbs of Henle via UTA2 ( Figure 24.15 ). Those processes are highly evolved in the outer medullary inner stripe where DVR and AVR are closely positioned in vascular bundles. Many water-conserving species also incorporate UTA2 expressing thin limbs of Henle within or on the periphery of vascular bundles ( Figure 24.9 ). Interesting insights into function have been obtained from the study of UTB-null mice. UTB deficiency results in reduced urinary concentrating ability, reduced urea clearance, and an increased plasma urea concentration. In contrast to wild-type mice, infusion of urea into UTB-null animals fails to enhance urinary concentrating ability. Acute regulation of UTB by vasopressin or other factors has not been demonstrated. In contrast to upregulation of UTA transporters, chronic vasopressin treatment may reduce UTB expression. UTB expression in the renal medulla increases during osmotic diuresis induced by urea, but not NaCl or glucose infusion. In contrast, UTB expression is depressed by ureteral obstruction, lithium treatment, potassium deficiency, and cyclosporine toxicity.

UTB expression in RBCs should limit AQP1-mediated water transport and the associated osmotic shrinking and swelling that would otherwise accompany RBC transit through the medulla. Macey and Yousef proposed that this might prevent osmotic lysis. Against this hypothesis is the finding that humans devoid of Kidd antigen have mildly depressed urinary concentrating ability, but no hemolytic anemia. It seems most likely that RBC expression of UTB serves to increase the overall mass of urea that is efficiently recycled from the AVR and DVR lumens in the renal inner medulla.

Transport of Macromolecules in the Cortex and Medulla

It is generally accepted that lymphatics are sparse in the outer medulla, and absent from the inner medulla. It has long been recognized that proteins permeate the walls of capillaries to be drained by lymphatics and returned to the systemic circulation. Given the absence of lymphatics in the inner medulla, the mechanisms that regulate interstitial oncotic pressure and protein trafficking through the interstitium have been enigmatic. Early studies led to the conclusion that a large extravascular pool of albumin is present within the medulla. Leakage of fluorescent albumin and Evans blue dye-labeled albumin into the medullary interstitium were observed. Ultrastructural studies with horseradish peroxidase (molecular radius 50 Å), catalase (elliptical molecule, 240,000 Da, major axis 240 Å), and ferritin (spherical molecule, 500,000 D, 110 Å) demonstrated that these markers can cross the fenestrations of cortical peritubular cortical capillaries and medullary AVR.

Measurements of albumin transport rates across the DVR and AVR walls have been technically limited. Using molecular sieving of Texas red-labeled albumin, Turner estimated the reflection coefficient of the DVR wall to albumin to be 0.89 (not significantly different from unity). In separate studies with different methods, the reflection coefficient of the AVR wall to albumin was estimated to be 0.7 and 0.78. No reliable measurements of the diffusional permeability of either the DVR or AVR wall to albumin exist. Attempts have been made to determine Starling forces within the medullary interstitium through direct measurement of interstitial protein concentration. Using a differential centrifugation technique, MacPhee and Michel obtained a mean value of 0.9 g/dl. By an alternative approach, interstitial protein concentrations of 4 to 6 g/dl were predicted and interstitial hydraulic pressures in the range of 5 to 10 mmHg were found.

Whatever the concentration of albumin in the medullary interstitium, the fundamental question remains, in the absence of lymphatics, how is medullary interstitial protein deposited and cleared by the microcirculation? Protein transport into the AVR lumen by convective influx is the most likely answer. Michel pointed out that molecular sieving at the AVR wall would indicate convective movement of protein into the AVR lumen, were it not for continuous deposition of protein-free fluid by medullary nephrons. The plausibility of convective protein uptake is also supported by the finding that papillary AVR withstand an inwardly-directed hydraulic pressure without collapsing. Pinter and colleagues have suggested that the combined effects of negative charge exclusion resulting from compartmentation of hyaluron and albumin, Donnan equilibrium, and hydrostatic pressure variation from ureteral contractions provide key driving forces for fluid movements and urinary concentration.