Renal Physiology

Concentration of solutes in the renal inner medulla: interstitial hyaluronan as a mechano-osmotic transducer

Mark A. Knepper, Gerald M. Saidel, Vincent C. Hascall, Terry Dwyer

This article has a correction. Please see:


Although the concentrating process in the renal outer medulla is well understood, the concentrating mechanism in the renal inner medulla remains an enigma. The purposes of this review are fourfold. 1) We summarize a theoretical basis for classifying all possible steady-state inner medullary countercurrent concentrating mechanisms based on mass balance principles.2) We review the major hypotheses that have been proposed to explain the axial osmolality gradient in the interstitium of the renal inner medulla. 3) We summarize and expand on the Schmidt-Nielsen hypothesis that the contractions of the renal pelvocalyceal wall may provide an important energy source for concentration in the inner medulla. 4) We discuss the special properties of hyaluronan, a glycosaminoglycan that is the chief component of a gel-like renal inner medullary interstitial matrix, which may allow it to function as a mechano-osmotic transducer, converting energy from the contractions of the pelvic wall to an axial osmolality gradient in the medulla. These considerations set the stage for renewed experimental investigation of the urinary concentrating process and a new generation of mathematical models of the renal concentrating mechanism, which treat the inner medullary interstitium as a viscoelastic system rather than a purely hydraulic system.

  • glycosaminoglycans
  • renal pelvis
  • vasopressin
  • hyaluronan

in states of fluid deprivation or nonrenal water loss, the kidney can conserve water while maintaining excretion of solutes. It does this by concentrating the solutes in the urine to osmolalities that markedly exceed the osmolality of plasma. A large number of studies, exemplified by the data shown in Fig.1, have demonstrated that the urinary concentrating process is associated with the generation of a corticomedullary osmolality gradient in the medullary tissue, oriented with the maximum osmolality in the deepest part of the inner medulla, i.e., at the papillary tip. The classic micropuncture studies of Gottschalk and Mylle (17) have established that the medullary hypertonicity is due to solute accumulation in all structures in the medulla, including loops of Henle, vasculature, and collecting ducts. The high medullary interstitial osmolality provides a driving force for osmotic water flow across the collecting ducts, which are rendered permeable to water through the action of vasopressin (33). The high water permeability allows osmotic equilibration of urine with the medullary interstitial fluid.

Fig. 1.

Osmolality gradient in renal medullary tissue of antidiuretic rabbit. Measurements were made by vapor pressure osmometry of slices from different levels of rabbit kidney after quick freezing. The figure is drawn from data from Ref. 32.

In 1959, Kuhn and Ramel (43) proposed a model to explain concentration of solutes in the renal medulla based on countercurrent amplification of a small osmotic difference between the ascending limb and the descending limb of Henle's loop, resulting from active solute transport out of the ascending limb. Their version of Hargitay and Kuhn's (22) countercurrent multiplier hypothesis has become generally accepted as the mode of solute accumulation in the renal outer medulla and is now supported by extensive experimental evidence (reviewed in Ref. 55). The key evidence was the demonstration that the thick ascending limb of Henle's loop is capable of a high rate of active NaCl transport out of the lumen, which results in luminal dilution owing to the low osmotic water permeability of this segment (3, 67).

Thus renal physiologists have developed a good understanding of the process that concentrates solutes in the renal outer medulla. The same cannot be said for the renal inner medulla, however. The ascending portion of Henle's loop (the thin ascending limb) has been shown to have extremely limited, if any, capacity for active transport in the inner medulla (25, 26, 53, 54, 59, 80). Therefore, in the inner medulla there is no energy source for a classic Kuhn-Ramel countercurrent multiplier, and other explanations must be sought for the medullary osmolality gradient in the renal inner medulla.

The objectives of this paper are 1) to summarize a simple theoretical scheme for classification of all possible steady-state countercurrent concentrating models in the inner medulla; 2) to review proposed steady-state models for concentration of solutes in the inner medullary interstitium; 3) to readdress the Schmidt-Nielsen hypothesis that energy from smooth muscle contractions of the pelvocalyceal wall is responsible for concentration of solutes in the inner medulla; and 4) to discuss the possible role of inner medullary interstitial hyaluronan as a mechano-osmotic energy transducer converting mechanical energy of renal pelvic contractions to axial osmolality gradients in the inner medulla.


Knepper and Stephenson (37) and Knepper et al. (34) developed a mathematical analysis of concentrating processes in the renal inner medulla that allows classification of all possible steady-state countercurrent concentrating models based on mass balance requirements. The full mathematical analysis will not be repeated here but is summarized concisely in appendix . This analysis assumes that solutions exhibit ideal behavior and that chemical reactions have negligible effects. Possible repercussions of these assumptions will be considered in later sections. We discuss the approach and the principles that derive from the analysis in the following paragraphs.

Figure 2 A is a diagram of a unipapillate kidney typical of a rat, rabbit, or mouse. It illustrates the relative positions of the three major regions of the kidney: the cortex, the outer medulla, and the inner medulla. The deepest portion of the inner medulla is a tapering structure, the papilla, whose tip is the site of exit of urine formed by the kidney. After exiting the kidney at the papillary tip, this urine is carried downward to the urinary bladder via the ureter. To analyze the processes responsible for generation of the osmotic gradient in the inner medulla, we apply a “control volume” for mass balance (Fig. 2 B), which creates a boundary to allow us to account for flows into and out of a portion of the inner medulla. The lower end of this control volume is defined to be at or just beyond the papillary tip. The upper end of the control volume is arbitrary; it can be drawn at any level of the inner medulla, for example, as defined by the solid line shown in Fig.2 B or by the horizontal dashed line just below it. The analysis that we present applies to all such control volumes. Figure2 C shows the same control volume identifying all of the relevant flows into and out of it. Entering flows include those in the descending vasa recta, the descending limb of Henle's loop, and the collecting ducts. Exiting flows include those in the ascending vasa recta, the ascending limbs of Henle's loop, and the final urinary flow exiting the papillary tip. At steady state, the flow of water, NaCl, and urea into the control volume must exactly equal the flows out of the control volume.

Fig. 2.

Structural basis of steady-state mass balance analysis of renal inner medulla of a unipapillate kidney. A: diagrammatic representation of kidney structure. See text for description. B: cutaway view showing a “control volume” for analysis of mass balance defined by the heavy solid rectangle. A series of such control volumes can be defined by moving the upper border of the control volume upward or downward. An alternative choice for the upper border is illustrated by the horizontal dashed line. For steady-state operation of the inner medulla, mass balance must be maintained for all such control volumes. C: enumeration of the individual flows into and out of an inner medullary control volume represented by the rectangle. Each flow stream is the aggregate of flows in all individual structures of a given type. For example, if the upper border of the control volume is assumed to be at the inner-outer medullary junction, the descending limb stream is the aggregate of 11,000 individual descending limbs (35). Asc., ascending; Desc., descending.

Figure 3 is a detailed view of the control volume for a portion of the inner medulla defining the terminology used. Here, the subscripts have been modified to indicate explicitly the structure being considered [e.g., descending vasa recta (DV); ascending vasa recta (AV); descending limb (DL); ascending limb of Henle's loop (AL); collecting duct (CD); final urine (U), peritubular interstitium (P)]. Total solute concentrations are represented by Cj, where the subscriptj designates the structure. Volume flow rates are represented by Qj. The products Cj · Qjrepresent the total solute flow rates into and out of the control volume. The total solute concentrations in the interstitium are given by the Cj terms. Using this terminology, the mass-blance equation (appendix , Eq. EA7 ) can be written in terms of the individual structures involved. This equation can be arranged so that the left-hand side expresses the interstitial total solute gradient from a given point (point x) along the inner medulla to the papillary tip L as followsCP(L)CP(x)=[QDV(x)/Qu][CDV(x)CP(x)]+[QAV(x)/Qu][CAV(x)CP(x)]+[QDL(x)/Qu][CDL(x)CP(x)]+[QAL(x)/Qu][CAL(x)CP(x)]+[QCD(x)/Qu][CCD(x)CP(x)] Equation 1To have a positive osmotic gradient between point x and the papillary tip [CP(L) − CP(x) > 0], at least one of the terms on the right-hand side must be positive. Each of these terms consists of a normalized flow multiplied by a total solute concentration difference (“osmolality difference”) across a given structure. Such a transverse osmolality difference has been referred to in the physiological literature as a “single effect,” a literal translation of the German term “Einzeleffekt” (22,43). The normalized flow is the absolute flow divided by the final urinary flow. The flow represents an aggregate of all flows for a given structure; e.g., QAL(x) is the sum of flows in all individual ascending limbs at level x. Each total solute concentration difference in the equation is a potential single effect, which could account for the positive gradient in the medulla.

Fig. 3.

Control volume for analysis of mass balance requirement for steady-state countercurrent multiplier mechanisms in the renal medulla. See text for definiton of terms.

The direction of the concentration difference (single effect) that is necessary for medullary interstitial concentration is dependent on the direction of flow in the structure. The flows oriented in the direction of the papillary tip (DV, DL, and CD) are positive and therefore require a positive value of Ci(x) − CP(x). The flows oriented away from the papillary tip (AV and AL) are negative and therefore require a negative value of Ci(x) − CP(x) to obtain a positive axial gradient. For the ascending limb, the requirement for a single effect is [CAL(x) − CP(x)] < 0, as is the case in the outer medulla due to active NaCl transport out of the thick ascending limb. Table 1summarizes all possible single effects that could account for a positive axial interstitial gradient in the inner medulla for steady-state operation. Potential concentrating models can therefore be analyzed on the basis of their ability to generate one or more of the required single effects indicated in Table 1. In summary, steady-state concentrating models must dilute the ascending limb of Henle or the ascending vasa recta relative to the surrounding interstitium or, alternatively, must concentrate the descending limb of Henle, the descending vasa recta, or the collecting duct relative to the surrounding interstitium. In the next section, we discuss some of the models that have been proposed in the context of the mass balance requirements summarized in Table 1.

View this table:
Table 1.

Classification of steady-state inner medullary concentrating models based on structure responsible for concentrating “single effect”


Single Effect in the Thin Ascending Limb of Henle

As noted in Table 1, a positive axial osmolality gradient in the inner medulla could be generated as a result of any process that dilutes the lumen of the thin ascending limb relative to the interstitium. The possibility that the thin ascending limb functions like the thick ascending limb to dilute its lumen relative to the interstitium by active NaCl transport has been ruled out, as discussed above. It has been proposed that urea may be actively reabsorbed from the thin ascending limb (45), although this hypothesis lacks experimental verification.

A model by which the luminal fluid in the thin ascending limb could be diluted by purely passive means has been proposed by Kokko and Rector (39) and by Stephenson (77) (Fig.4). This model assumes that the luminal fluid at the bend of the loop contains NaCl as the predominant solute and that the inner medullary interstitium contains a fluid in which urea is the predominant solute. It has been noted that the permeability to NaCl is higher than the permeability to urea in the thin ascending limb (25). These permeability characteristics predict that NaCl would escape the lumen more rapidly than urea would enter, resulting in passive dilution, a prediction that was bourne out by perfused tubule experiments in vitro (25). Although this model appeared promising at first, thorough quantitative analysis did not support an important contribution of this process to the generation of an inner medullary osmolality gradient (5, 6, 46, 81, 83,92). Furthermore, a physiological analysis on the basis of the known permeability properties, medullary solute concentrations, and flow rates in medullary structures has led to the conclusion that the Kokko-Rector-Stephenson passive model could account for only a modest axial osmolality gradient in the inner medulla (55). One important discrepancy between the experimental data and the requirements of the Kokko-Rector-Stephenson model is the urea permeability of the ascending thin limb epithelium. Measurements in isolated perfused rodent thin ascending limbs yielded extremely high values, in the range 38–170 × 10−5 cm/s (7, 24), which is too high to permit sustained net luminal dilution along the length of the thin ascending limb. In addition, measurements of the urea permeability of the descending limb epithelium in the inner medulla demonstrated values too high to prevent substantial urea entry into the descending limb (46). It is beyond the scope of this article to analyze the full evidence in detail; see Masilamani et al. (55) for a thorough analysis of the feasibility of the Kokko-Rector-Stephenson model.

Fig. 4.

Proposed function of the thin ascending limb of Henle in the Kokko-Rector-Stephenson passive model for concentration of the inner medullary interstitium. The entering fluid and peritubular fluid are assumed to have the same osmolality (osm) but to have equal and opposite transepithelial gradients for urea and NaCl. If the thin ascending limb has the special properties indicated in the text (high NaCl permeability, low urea permeability, low water permeability), rapid NaCl efflux would occur without a balancing entry of urea, thus lowering luminal osmolality below that of the peritubular fluid. Measurements of urea permeability of the thin ascending limb have yielded relatively high values, seemingly ruling out the hypothesis (see text).

Although the existing literature raises considerable doubt about the view that the single effect for inner medullary concentration resides in the thin ascending limb, recent evidence from studies of gene knockouts of the ClC-K1 chloride channel in mice emphasizes that the thin ascending limb is nonetheless important in the inner medullary concentrating mechanism (1, 56). ClC-K1 is expressed exclusively in the thin ascending limb, where it is responsible for extraordinarily high chloride permeabilities in that segment (85,86). The knockout mice exhibited a severe concentrating defect and a failure to substantially concentrate the inner medullary interstitium in association with a low chloride permeability in its thin ascending limbs. The basis of the defect can be understood fromEq. EA1 in appendix . As can be appreciated, if the transepithelial osmolality gradient across the thin ascending limb were oriented lumen > interstitium, this would provide a negative term in the equation calculating the axial osmolality gradient (compare with single-effect condition, Table 1). This would create essentially a “negative single effect” as fluid flowed upward from the highly concentrated papillary tip to the less-concentrated outer medulla. Thus failure of osmotic equilibration across the thin ascending limb would decrease the axial gradient that could be generated in the inner medulla by any mechanism. Therefore, high permeabilities to NaCl and urea in the thin ascending limb are extremely important as a means of preventing dissipation of the inner medullary solutes by the upward flow in thin ascending limbs as illustrated by the ClC-K1 knockout studies.

Single Effect in the Descending Limb of Henle

As noted in Table 1, a positive axial osmolality gradient in the inner medulla could be generated as a result of any process that increases the osmolality of the luminal fluid in the thin descending limb relative to the interstitium. If active transport of solute in this segment were to be implicated in generation of a single effect, the direction of transport would have to be in the secretory direction, i.e., into the lumen. Kriz and colleagues (41, 42) have proposed such a solute secretory mechanism as part of a “cascade model” of inner medullary concentration, but there is thus far no evidence to support the presence of active solute transport into the thin descending limb. Bonventre and Lechene (2) have also presented a similar concept. They suggested that the tubule fluid of the long descending limbs in the outer medulla may be hypertonic to the interstitium of the upper part of the inner medulla because of selective interaction with the interstitial subregion that surrounds the thick ascending limbs in the outer medulla. Mathematical modeling studies by Lory (48) have predicted that even with a substantial rate of solute transport into the descending limb, other concomitant processes would decrease the axial osmolality gradient in the renal medulla. Apparently, the extremely high water permeability of the descending limb would make it impossible for the thin limb to sustain the required transepithelial osmolality difference. The high water permeability of this segment owes to the abundant expression of the aquaporin-1 water channel in the plasma membranes of the thin descending limb cells (8, 50).

Given the high water permeability of the thin descending limb, it has also been proposed that a single effect (luminal osmolality > interstitial osmolality) could be generated as a result of unequal osmotic reflection coefficients for urea and NaCl (34). The urea concentration in the interstitium is higher than that in the descending limb lumen, whereas the NaCl concentration in the lumen is higher than that in the interstitium (15, 29, 52). Under these circumstances, the urea gradient would tend to drive water out of the lumen and the NaCl gradient would tend to drive water inward. If the reflection coefficient for NaCl were lower than that for urea, osmotic equilibration would occur with a higher NaCl gradient than the opposing urea gradient, resulting in an osmolality that is higher in the lumen than the interstitium. That is, a single effect would be generated (Table 1). This hypothesis has not been experimentally tested for thin descending limbs from the inner medulla. However, the hypothesis seems somewhat questionable based on characterization of the aquaporin-1 water channel, the main pathway for water movement across the thin descending limb. Fundamentally, a reflection coefficient of <1 for a given solute requires that the water pathway across the barrier membranes be permeable to that solute (30). Measurements of solute permeability of aquaporin-1 heterologously expressed in Xenopus laevis oocytes or reconstituted into artificial lipid vesicles indicate that the urea and NaCl permeabilities are extremely low (64, 95). However, passive cation fluxes associated with aquaporin-1 have been reported inX. laevis oocyte expression studies, and these fluxes have been noted to be increased by cAMP treatment (94). Nevertheless, the chloride permeability remained very low, seemingly ruling out substantial net penetration of NaCl through aquaporin-1. Because the reservations to this model are based on theoretical considerations only, direct measurements in isolated perfused inner medullary descending limbs will be necessary to rule out the hypothesis with certainty.

Single Effect in the Collecting Duct

As summarized in Table 1, a single effect accounting for an axial osmolality gradient in the inner medulla could theoretically be generated in the collecting duct if the luminal osmolality were maintained greater than that of the surrounding interstitium throughout most of the inner medulla. Wexler et al. (93) have proposed one means by which this could happen. Specifically, an extremely hyperosmotic fluid may be generated in the outer medullary collecting duct and delivered to the inner medullary collecting duct (IMCD). Based on the prior studies of Lemley and Kriz (47), it has been concluded that the outer medullary collecting ducts are segregated with the thick ascending limbs in the outer medulla. Theoretically, the rapid active NaCl transport from the thick ascending limbs would concentrate the interstitium adjacent to the collecting ducts to a level that greatly exceeds the average osmolality of the outer medullary tissue. This would raise the osmolality to a high level in the collecting duct lumen, and this highly concentrated fluid would enter the IMCDs, providing a single effect (luminal osmolality > interstitial osmolality). A mathematical model devised by Wexler and colleagues (93) showed that such a process could result in concentration of the inner medullary interstitium, but this model required a special condition: the osmotic water permeability of the initial portion of the IMCD was required to be very low to prevent the high luminal osmolality from being dissipated by the water secretion that would otherwise occur. Experimental studies by Han and colleagues (21) did not confirm that key assumption and instead found a high osmotic water permeability in the initial IMCD, apparently ruling out the model (21). Subsequent papers by Wang and colleagues (87,88) suggested that the requirement for a low water permeability in the initial IMCD could be relaxed somewhat if a high rate of rapid active NaCl transport occurs out of the initial IMCD. However, based on further analysis using a detailed three-dimensional model of the medullary concentrating process, Thomas and Wexler (83) concluded that

if realistic values of urea permeability in the inner medullary descending limbs and water permeability in the upper inner medullary section of the collecting ducts are taken into account, even a model including the three-dimensional vascular bundle structures fails to explain the experimentally observed inner medullary osmolality gradient.

A different model for generation of a single effect (luminal osmolality > interstitial osmolality) in the IMCD, based on a difference in reflection coefficients for urea and NaCl, has been proposed several times (2, 5, 6, 20, 27, 65, 68). The principle is similar to that proposed for generation of a single effect in the thin descending limb as discussed above, except that the directions of the transepithelial urea and NaCl gradients in the IMCD are opposite those seen in the descending limb. Therefore, the proposed model depends on the assumption that in the IMCD, the reflection coefficient for urea is much lower than that for NaCl. Indeed, early measurements of reflection coefficients in the IMCD seemed consistent with this assumption [see Morgan and Berliner (58) and Imai et al. (27), for example]. However, subsequently the transport proteins responsible for water transport [aquaporin-2 (14)] and urea transport [UT-A1 (74)] across the apical plasma membrane of the IMCD have been identified by molecular cloning, providing direct evidence for independent, highly selective transport pathways for water and urea. These results provided no evidence for a shared pathway for water and urea transport as required for a true reflection coefficient of <1 (30). Indeed, careful measurements in isolated perfused tubules demonstrated that the reflection coefficient for urea is virtually 1 and that the apparent low value of the reflection coefficient was due to rapid dissipation of imposed urea gradients by facilitated urea transport (9, 36). A mathematical analysis of transport of solutes and water across the IMCD indicates that the presence of unstirred layers (chiefly in the cytoplasm) can contribute to the presence of an apparent reflection coefficient for urea of <1 without having nonunity reflection coefficients for transport across the plasma membranes (19).

In general, models that depend on a single effect in the IMCD are at a theoretical disadvantage relative to models that depend on a single effect in the loop of Henle because of the low aggregate tubule fluid flow rate in collecting ducts relative to the loop of Henle. As can be seen in Eq. EA1 in appendix , the degree of countercurrent multiplication is directly proportional to both the normalized tubule fluid flow rate and the magnitude of the single effect.

Proposed Single-Effect Mechanisms in the Vasa Recta

The ascending vasa recta are lined by fenestrated endothelial cells (57), which presumably permit free exchange of solutes and water between the interstitium and the ascending vas rectum lumen. Consequently, the composition of the interstitial fluid has generally been assumed to be similar to that of the blood plasma in the ascending vas rectum. In contrast, the endothelium of the descending vasa recta is continuous, and transendothelial gradients are a possibility. Thus it is conceivable that the inner medullary concentrating process could be driven by generation of a single effect in the descending vasa recta, i.e., a process that maintains the osmolality of the lumen greater than that of surrounding interstitium (Table 1). The water permeability of the thin descending limbs is very high due to the expression of very high levels of the water channel aquaporin-1 in the plasma membranes of the endothelial cells (61). Therefore, steady-state models that depend on active or passive solute transport are unlikely to generate sustained osmolality gradients, as discussed above with regard to the descending limb of Henle's loop. Nonetheless, it is conceivable that a process based on a difference in reflection coefficients for NaCl and urea could contribute to the concentration of solutes in the inner medullary interstitium by increasing the luminal osmolality above that of the surrounding interstitium. The same reservations can be made about this model in the descending vasa recta as were made for the descending limb of Henle (see above).


Conventional models of the medullary concentrating process have assumed steady-state conditions, ideal solutions, and negligible solute production by chemical reactions (see appendix ). However, the appropriateness of these assumptions has been questioned. In the remainder of this review, we examine potential concentrating models based on reconsideration of these assumptions.

Possible Role of Solute Generation Via Chemical Reactions

In 1994, Jen and Stephenson (28) provided theoretical justification for the view that generation of some “external osmolyte” in the inner medulla could provide the driving force for the inner medullary concentrating process. In their formulation, an unspecified solute is assumed to be added de novo and continuously to the inner medullary interstitium. A subsequent mathematical modeling study by Thomas and Wexler (83) using a complex three-dimensional model of the renal medulla confirmed that addition of such a solute to the inner medullary interstitium could potentially explain the axial concentration gradient in the inner medulla by driving water efflux from the thin descending limb. This would concentrate NaCl in the descending limb, setting up a favorable gradient for NaCl efflux from the ascending limb and dilution of the ascending limb lumen relative to the interstitium. In effect, this external solute substitutes for urea in the Kokko-Rector-Stephenson passive model.

What could be the identity of this “external solute”? The model requires a chemical reaction that generates more osmotically active particles than it consumes. Thomas (82) has proposed that the external solute is lactate, which is generated by anaerobic glycolysis (the predominant means of ATP generation in the inner medulla) in the proportion of 2 lactate ions/glucose molecule consumed:glucose2lactate+2H+ The feasibility of the proposed model depends on the fate of the H+ ions that are generated. If the H+ ions titrate HCO 3 , they will remove two osmotically active particles (HCO 3 ions), resulting in a net disappearance of osmotically active particlesglucose+2HCO32lactate+2CO2 Because CO2 readily permeates lipid bilayers, it is unlikely to be osmotically effective. Alternatively, if the H+ ions titrate buffers other than HCO 3 , e.g., NH3, phosphate, and proteins with relatively neutral isoelectric points, a net generation of osmotically active particles can be expected.

In proposing that lactate generation provides a driving force for the inner medullary concentrating process, Thomas (82) raises critical questions about lactate transport in the inner medulla. For the model to explain the axial osmolality gradient in the inner medulla, the lactate generated would need to be transported out of the epithelial and endothelial cells in a polarized fashion, so as to generate osmotic differences across individual renal tubule segments (Table 1). Thomas proposes that lactate should be preferentially transported across the basolateral plasma membranes of all cell types.

Possible Role of Solution Nonideality in the Renal Medulla

Most models of the urinary concentrating mechanism have assumed that the tubule fluid and interstitial fluids in the renal medulla behave as ideal solutions. However, as pointed out by Wang et al. (89), renal medullary fluids may deviate substantially from ideality under antidiuretic conditions. Their preliminary calculations using a complex multinephron model of the renal medulla indicate that compared with ideal solution models, decreased activity coefficients for urea tend to increase predicted urinary osmolalities, while decreased activity coefficients for NaCl tend to decrease predicted urinary osmolalities. Such nonideal effects probably should be taken into consideration in any model of the urinary concentrating process.

Non-Steady-State, Periodic Models

The potential role of periodic contractions of the renal pelvic wall as a source of energy for the concentrating process in the inner medulla has been emphasized by Schmidt-Nielsen (70). The renal inner medulla is surrounded by the renal pelvocalyceal wall (Fig.2 A), a structure comprised chiefly of two thick smooth muscle layers (75). The pelvocalyceal wall undergoes intermittent contractions, which have been seen to compress the renal medullary parenchyma (71) (Fig.5). These compressions, occurring at a frequency of 15–40/min in rodents, have been seen to alter flow rates in tubule and vascular structures of the inner medulla (66,73). The frequency of the contractions is regulated via both sympathetic and parasympathetic inputs (12, 40). Thus pelvocalyceal wall contractions impart a periodic character to the function of the inner medulla that has been largely ignored in formal mathematical modeling studies of the urinary concentrating process. As pointed out by Schmidt-Nielsen (70), the pelvic wall contractions could provide an energy input to the concentrating process itself. In support of this view is the longstanding observation that disruption of the continuity of the pelvocalyceal wall markedly reduces the tonicity of the inner medullary tissue and urine (10, 16,60). Two studies have directly addressed this possibility with differing conclusions. Oliver et al. (60) tested the effects of paralyzing the upper portion of the ureter (which surrounds the papillary tip) in young rats and found no impairment of concentrating ability. In contrast, Schmidt-Nielsen et al. (72) found in hamsters that paralysis of the pelvic wall (or mechanical damage to the pelvic wall) significantly decreased concentrating ability. In the remainder of this article, we present a discussion of ways that periodic contractions of the renal pelvic wall could concentrate the inner medulla, including consideration of the role of hyaluronan in the interstitial matrix as a molecular mechano-osmotic transducer for the concentrating process.

Fig. 5.

Renal pelvic wall contractions in hamster kidney.A: back-illuminated image of renal papilla. Video clip showing contractions can be viewed at URL or through the WWW at URL (login: lkem; password: review). B: optical recordings of intensity of transmitted light during contractions of renal pelvic wall showing morphology and frequency of contractions. The contraction phase typically takes up 1/5 of the cycle and the relaxation phase 4/5 of the cycle. Adapted from Ref. 69 with permission.

Before a consideration of specific models, it is important to note that the formulation which defines possible single-effect mechanisms for the inner medulla given in appendix applies only to the steady state. Further work will be needed to extend the analysis to periodic and other non-steady-state conditions. Nevertheless, it is reasonable to assume that in the periodic case, single-effect conditions similar to those presented in Table 1 will apply. In particular, we assume that positive axial osmolality gradients will be generated when the contents of the descending limbs (or descending vasa recta) are concentrated relative to the surrounding interstitium or when the contents of the ascending limbs (or ascending vasa recta) are diluted relative to the surrounding interstitium.


Schmidt-Nielsen (70) was the first to emphasize the remarkable spongelike properties of the renal interstitial hyaluronan matrix and the potential role of hyaluronan in concentration of the medullary interstitium. In this section, we describe a concentrating model based on the view that the inner medullary interstitium consists of a semisolid, viscoelastic hyaluronan gel rather than being a freely flowing aqueous compartment. As outlined in detail below, it is proposed that the hyaluronan matrix can store the mechanical energy from the pelvic contractions by direct mechanical compression without the need to generate high hydrostatic pressures and can utilize this energy to lower interstitial pressure after completion of each contraction of the pelvic wall to drive water efflux from the descending limb of Henle. The latter process would increase the luminal osmolality to above that of the interstitium, thereby generating a single effect for concentration of urine. Before laying out this model, we summarize the necessary background regarding the biochemistry of hyaluronan and its physicochemical properties.

Hyaluronan (hyaluronic acid) is a member of a family of biomolecules called glycosaminoglycans (GAGs), which are all unbranched polysaccharide chains composed of repeating disaccharide units. Aside from hyaluronan, other mammalian GAGs include chondroitin sulfates, dermatan sulfate, keratan sulfate, heparan sulfate, and heparin. Hyaluronan differs from the other GAGs in that it is not generally covalently linked to proteins to form proteoglycans and is not sulfated (23). Furthermore, in contrast to the other GAGs that are synthesized in the Golgi apparatus, hyaluronan is produced at the plasma membrane by an integral membrane protein, hyaluronan synthase (HAS) (84, 90). Three mammalian HAS genes have been identified, namely, HAS1, HAS2, andHAS3. All three produce hyaluronan on the cytoplasmic side of the plasma membrane and transport it across the plasma membrane to the extracellular fluid. Thus hyaluronan secretion does not directly involve vesicular trafficking, in contrast to most other types of secreted biomolecules. Because of the importance of GAGs in the structure of connective tissues, such as cartilage, bone, synovial fluid, intervertebral disks, tendon, skin, and cornea, the physicochemical properties of these substances have been thoroughly characterized (11).

Several studies have demonstrated that hyaluronan is highly abundant in the interstitium of the renal inner medulla in contrast to the low amounts seen in other regions of the kidney (4, 13, 18). Figure 6 illustrates the high level of hyaluronan accumulation in the rat inner medulla as revealed by Alcian blue staining. Other GAGs are present in the inner medulla in much lower amounts. The hyaluronan in the inner medulla is believed to be produced by a specialized interstitial cell (the so-called type 1 interstitial cell) that forms characteristic “bridges” between the thin limbs of Henle and vasa recta (62).

Fig. 6.

Alcian blue staining of normal rat kidney revealing distribution of hyaluronan in inner medulla. Bar = 2 mm.

Figure 6 illustrates that the hyaluronan-laden inner medulla is contained within the renal pelvic wall with its thick smooth muscle lamina. The compression of hyaluronan in the medullary interstitium by the peristaltic contractions of the pelvic wall can hypothetically serve to generate a single effect for inner medullary concentration in two ways: 1) by lowering the osmolality of the surrounding interstitial fluid; and 2) by storing mechanical energy, which when released can create forces that drive water absorption from the descending limb of Henle. We consider these two mechanisms in turn.

Hyaluronan Contraction May Lower Local Osmolality

Hyaluronan is a large, unbranched polysaccharide molecule composed of repeating glucuronic acid/N-acetylglucosamine disaccharide subunits (Fig.7 A).1It is a polyanion, owing to the carboxylate groups of the glucuronic acid subunits. It is a huge molecule, typically with a molecular mass in the range 1,000–10,000 kDa. It is strongly hydrophilic and adopts highly expanded, stiffened random-coil conformations that occupy a huge volume relative to their mass. In solutions of physiological ionic strengths, the domains of individual molecules begin to overlap at low concentrations (<5 mg/ml). The hydrodynamic domains are readily compressed when concentrated under a mechanical load and expand when the compressive force is removed. Thus the inner medullary interstitium can be visualized as being composed of a compressible, viscoelastic hyaluronan matrix. The extended state of hyaluronan owes partly to repulsive electrostatic forces exerted by neighboring COOgroups, which maximize the distance between neighboring negative charges (Fig. 7 B), and partly by the constraints of the glycosidic bonds that prefer somewhat extended conformations. This creates a swelling pressure (turgor) that allows the hyaluronan matrix to generate an elastic-like force (resiliance) that resists compression. When HA is compressed, as may occur in a meniscus in the knee joint under load-bearing conditions, the repulsive force of neighboring COO groups is overcome in part by immobilization (or “condensation”) of cations (chiefly Na+), forming a localized crystalloid structure (Fig.7 C). Thus compression of a hyaluronan gel results in a lowering of the local Na+ ion activity in the gel. In aqueous solutions in Donnan equilibrium with the gel, one can predict a decrease in the NaCl concentration secondarily to the compression-induced reduction in Na+ activity within the gel. Thus the free fluid that can be expressed from such a gel would have a lower total solute concentration than that of the gel as a whole. Such an effect could be important in the renal inner medulla when the force of pelvocalyceal contractions compresses the inner medullary interstitial matrix. The slightly hypotonic fluid expressed from the interstitial matrix would tend to escape the inner medulla via the ascending vasa recta, which is the only structure that remains open during the compressive phase of the contraction cycle (49). Thus an ascending stream (the ascending vasa recta) would have a lower total solute concentration than the interstitium as a whole and therefore this would create a single effect for medullary concentration (Table 1).

Fig. 7.

Structure of hyaluronan. A: disaccharide subunit. B: extended polyanion. Hyaluronan is a linear polymer of the disaccharide shown in A, with a molecular mass in the range 1,000–10,000 kDa. Polymer tends to remain in the extended state because of repulsion of negative charges of carboxylate groups. C: when the hyaluronan polyanion is compressed, free cations are sequestered.

Conversion of Mechanical Energy to Chemical Potential Energy Via the Viscoelastic Properties of Hyaluronan

During the relaxation phase of the pelvocalyceal contraction-relaxation cycle, two additional processes may contribute to urinary concentration through creation of a single effect (luminal osmolality > interstitial osmolality) across the thin descending limb epithelium. Both processes result from relaxation of the compressed hyaluronan matrix: 1) water would be absorbed from the descending limb as a result of a decrease in the hydrostatic pressure in the medullary interstitium; and 2) water would be absorbed from the descending limb as a result of elastic forces exerted directly by the expanding medullary interstitial matrix. The relevant forces have been described by Maroudas (51) in the analysis of water transport from articular cartilage, a glycosaminoglycan-filled tissue similar in properties to the hyaluronan-filled inner medullary interstitium. Fundamentally, the flux of water into and out of cartilage (or the inner medullary interstitium) may be viewed as being driven by three forces, expressed as pressure differences: ΔPosmotic, ΔPhydrostatic, and ΔPelastic. In the context of forces determining water transport between the thin descending limb and the hyaluronan matrix of the inner medulla, ΔPosmotic represents the osmolality difference between the lumen and the interstitial matrix, ΔPhydrostaticrepresents the hydrostatic pressure difference between the lumen and the interstitial matrix, and ΔPelastic represents the force exerted due to the elastic deformation of the interstitial matrix (given here as an equivalent pressure difference). According to this formulation, during the relaxation phase after passage of the pelvic peristaltic wave, elastic forces from expansion of the compressed hyaluronan would increase water transport in two ways: 1) ΔPelastic could directly draw water out of the descending limb (and other water-permeable structures); and 2) the tendency to interstitial expansion due to the relaxation of the compressed hyaluronan may lower the interstitial pressure below ambient pressure levels to produce a hydrostatic pressure difference ΔPhydrostatic, increasing water withdrawal from the descending limb and other water-permeable structures. The tendency of the pressure drop in the interstitium to cause cavitation would be countered by the gel structure. In the inner medulla, the flow of water driven by the sum of elastic and hydrostatic pressure forces (ΔPelastic + ΔPhydrostatic ) would concentrate the lumen of the descending limb relative to the interstitium. As water flows out of the descending limb of Henle, the limiting condition of no water flow is approached where ΔPosmotic = ΔPelastic + ΔPhydrostatic. Here, ΔPosmoticrepresents a limiting single-effect value.

It is possible that the sum ΔPelastic + ΔPhydrostatic may be much larger than 1 atm, although measurements of this force are not presently available. It is important to reemphasize that this represents a very low pressure in the interstitium rather than a very high pressure in the renal tubule relative to ambient pressures. The fall in hydrostatic pressure in the interstitial matrix would be expected to be bounded, if one assumes that “absolute negative” hydrostatic pressures are an impossibility, so that ΔPhydrostatic would not exceed 1 atm. Nevertheless, negative absolute pressures have been reported, for example, in the xylem of trees as a result of transpiration (70). These pressures are believed to furnish the driving force for the flow of water upward from the roots to the tree tops of large deciduous trees, overcoming the weight of a 200-ft column of water. Xylem pressures of −5 to −6 atm relative to ambient pressure have been reported (63). For values of ΔPelastic + ΔPhydrostatic ranging from −1 to −6 atm, the value of the single-effect osmolality difference across the thin descending limb would range from 40 to 240 mosmol/kgH2O [ΔCosm = (ΔPelastic + ΔPhydrostatic)/π · RT]. As described in appendix , this would give a urinary osmolality in the range from 1,416 to 4,000 mosmol/kgH2O, spanning the value of maximal urinary osmolality measured in normal rats (2,900 mosmol/kgH2O) (31).

In summary, the chief new concept presented in this review is that the inner medullary interstitium might best be modeled as a viscoelastic system with stress-strain properties, rather than a purely hydraulic system. This necessitates consideration of force terms other than hydrostatic pressure and osmotic pressure. The main additional force term that we add to the analysis is elastic force. During inner medullary compression resulting from the contraction of the pelvic wall, the compression of the hyaluronan matrix stores some of the mechanical energy generated from the smooth muscle contraction. This compression would not require an increase in hydrostatic pressure but would simply require a direct mechanical compression of the hyaluronan matrix as one would compress a steel spring. After passage of a peristaltic wave, the compressed hyaluronan will tend to spring back from its compressed state, exerting an elastic force and lowering interstitial pressure, thereby driving water from the descending limb and other water permeable structures. The water efflux would concentrate solutes in the tubule lumina. This would complete an energy conversion starting with ATP hydrolysis in smooth muscle cells of the pelvic wall, leading to compression of the hyaluronan in the medullary interstitium and then to an increase in electrochemical potential due to concentration of solutes in the tubule lumina. We have analyzed this process here only with regard to mass balance requirements. Clearly, further theoretical and experimental analysis is required to evaluate the feasibility of these proposed energy transfers purely on the basis of energy balance, as done previously for steady-state systems (78, 79, 91). The single effect generated from this process could add to single effects from other processes, e.g., lactate generation in the renal medulla, to concentrate the urine.

An important question that must be addressed experimentally is whether the rate of energy generation by ATP hydrolysis and contraction of the pelvic wall is sufficient to account for the energy input needed to concentrate the collecting duct urine as it flows along the inner medullary axis. An additional question concerns the stress-stain properties of the renal papilla and whether the modulus of elasticity is sufficient to mediate the proposed mechanical energy transduction. Finally, an important experimental question is the degree to which the basement membrane of the thin limbs of Henle can withstand hypothetical transepithelial pressure differences of 1 atm or more without undergoing permanent deformation. Perhaps the small radius of these tubules plays an important role in limiting the wall tension needed to counter such pressure forces according to the Law of Laplace (ΔP = T/r).


The identification of the process responsible for concentration of solutes in the interstitium of the inner medulla has been elusive. The lack of definition of these mechanisms undoubtedly owes to the technical difficulty of studying processes in the intact renal medulla without disrupting these processes. In recent years, interest in investigation of this problem has flagged as renal physiologists have turned their attention to individual genes and proteins, focusing on the molecular aspects of transport regulation and especially on processes that are amenable to study in cell culture. Nevertheless, the purely integrative question of how the inner medullary interstitium is concentrated remains as important as ever. This review has been presented with the idea of stimulating further work on the problem. By proposing specific hypotheses involving specific genes and gene products, e.g., the hyalurononan synthase (HAS) genes and hyaluronan, we hope to stimulate investigators to reexamine this problem with the tools of 21st-century physiology.

For example, it may be possible to use transgenic and gene knockout technology to address critical elements of the model such as1) targeted/conditional knockouts of the HAS2gene in the renal inner medulla; 2) targeted deletion of interstitial cells of the renal inner medulla, which produce the interstitial HA; and 3) targeted deletion of contractile proteins of the pelvic wall.


The authors gratefully acknowledge the career and contributions of Bodil Schmidt-Nielsen; many of the concepts presented in this review originated in her work.


  • The authors thank Lisa Hennegar for assistance with the preparation of Fig. 6 (Alcian blue histology).

  • Funding for this work was derived from the intramural budget of the National Heart, Lung, and Blood Institute (Z01-HL-01282-KE to M. A. Knepper) and from extramural National Heart, Lung, and Blood Institute Grant HL-51971 (to T. Dwyer).

  • 1 For a detailed description of the structure and properties of hyaluronan, the reader is referred to Ref.44 or the WWW site at the URL

  • Address for reprint requests and other correspondence: M. A. Knepper, National Institutes of Health, Bldg. 10, Rm. 6N260, 10 Center Dr., MSC 1603, Bethesda, MD 20892-1603 (E-mail:knep{at}

  • 10.1152/ajprenal.00067.2002


Derivation of General Mass Balance Equation for Renal Inner Medulla for Steady State

If we view the renal medulla as consisting of parallel tubes (renal tubule segments and vasa recta), oriented in the xdirection, we can define the following terms:Qi(x)=net axial volume flow in theith tube Cik(x)  = concentration of the kth solute in the i th tube at medullary level x

Qi(x)  Cik(x)  = net axial flow of the k th solute in the i th tube at medullary level x

The flows are assumed to be positive if oriented toward the renal papillary tip and negative if oriented toward the cortex (see Fig. 2 for definition of structures and control volumes).

Considering a control volume bounded by x 1 at the top and x 2 at thebottom, mass balance equations can be written for water and individual solutes. Assuming that there are no chemical reactions in the system that create or destroy any of the solutes, and that the fluid density is everywhere equal, water balance is given byiQi(x1)=iQi(x2) Equation A1and mass balances for each solute k are given byiQi(x1)Cik(x1)=iQi(x2)Cik(x2) Equation A2Because we are interested specifically in concentration of the final urine, we consider the balance where the bottomboundary of the control volume x 2 is at pointL, the papillary tip. In this case, there is a single flow stream crossing the bottom boundary, viz., the final urine. Defining Qu as the final urinary volume flow rate and Ck u as the concentration of solutek in the final urine, Eqs. EA1 and EA2 can be rewritteniQi(x1)=Qu Equation A3andiQi(x1)Cik(x1)=QuCku Equation A4For this analysis, we wish to consider the total solute concentration rather than the concentrations of the individual solutes. Consequently, we define Ci M(x) as the total solute concentration in the ith tube at axial levelx, which is given byCiM(x)=kCik(x) The total solute concentration in the final urine is defined as CMu . Using these terms, a mass balance for total solutes can be derived from Eq. EA4 , yieldingiQi(x1)CiM(x1)=QuCMu Equation A5The analysis so far does not take into account the peritubular interstitial composition. We define CPM(x) to be the peritubular interstitial total solute concentration at levelx. In doing so, we assume that the interstitial composition is not a function of position in a direction perpendicular to thex-axis. We can modify Eq. EA5 by arbitrarily subtracting the term Qu CPM(x 1) from both sides and rearranging (usingEq. EA3 to substitute for Qu on the left-hand side), givingiQi(x1)[CiM(x1)CPM(x1)]=Qu[CMuCPM(x1)] Equation A6If we assume that the collecting duct is in osmotic equilibrium with the medullary peritubular interstitium at the papillary tip (x = L), i.e., CPM(L) = CMu, then Eq. EA6 can be rearranged to give the axial total solute gradient from point x 1 to the papillary tipCPM(L)CPM(x1)=i[Qi(x1)/Qu][CiM(x1)CPM(x1)] Equation A7In other words, the osmotic gradient in the renal medullary interstitium (expressed on the left-hand side of the equation) is determined by the total solute concentration gradient across each structure multiplied by the normalized flow rate in that structure. The implications of this equation are described in the text.


Calculation of Maximal Urinary Osmolality That Could be Generated by Pelvic Contractions Based on Mechano-osmotic Energy Transduction by Hyaluronan

If we consider a rat with a mean single nephron GFR in long-loop nephrons of 45 nl/min excreting 0.25% of filtered water, we can calculate the multiplier factor [QDL(x)/Qu] for text Eq.1 , representing the ratio of total flow in all descending limbs at the inner-outer medullary junction to the final urinary flow. Of the 38,000 nephrons in a rat kidney (35), 29% have loops of Henle that extend into the inner medulla (76), giving a total of 11,000 long-looped nephrons. If we assume that two-thirds of the filtered water is absorbed in the proximal tubule (17), the aggregate flow out of the proximal tubules can be calculated to be 15 nl · min−1 · nephron−1× 11,000 nephrons = 165 μl/min. If we assume that the osmolality in the descending limbs increases from 300 to 900 mosmol/kgH2O in the outer medulla solely as a result of water abstraction (38), then the aggregate flow in descending limbs entering the inner medulla would be 165 μl/min × (300/900) = 55 μl/min. The final urinary flow (assuming a mean SNGFR of 45 nl/min) would be 45 nl · min−1 · nephron−1× 38,000 nephrons × 0.0025 = 4.28 μl/min. The flow ratio or multiplier factor [QDL(x)/Qu] is 55/4.28 = 12.9. This flow ratio would be the mean integrated flow ratio over the entire pelvic contraction cycle. For mean single effect values ranging from 40 to 240 mosmol/kgH2O, the axial inner medullary osmolality gradient calculated from textEq. 1 would range from 516 to 3,096 mosmol/kgH2O. This value would be added to the osmolality at the inner-outer junction (900 mosmol/kgH2O) to get a range of values for maximal urinary osmolality from 1,416 to 3,996 mosmol/kgH2O.

These calculations ignore single effects that might be generated in the descending vasa recta and collecting ducts by the same mechanism but also ignores the fact that dissipative terms in text Eq. 1 would tend to reduce the gradient generated.


View Abstract