INTRODUCTION

Top Abstract Introduction References

THE JUXTAGLOMERULAR APPARATUS (JGA) is a structure in the mammalian nephron where a specialized tubular epithelium, the macula densa (MD) cell plaque, makes regular contact with the glomerular arterioles at their entrance into the glomerulus. Interposed between the MD and the vascular cells are the interstitial cells of the extraglomerular mesangium. Thus the JGA is structured to permit an interaction between tubular epithelial cells and smooth muscle cells of the afferent arteriole, a site determining glomerular capillary pressure and glomerular filtration rate (GFR), and between epithelial cells and granular cells, the site of production and release of renin. This arrangement of cells seems to be a perfect example for Starling's assertion, expressed in his Principles of Human Physiology (123), that "in no other organ of the body are our views as to function so intimately dependent on our knowledge of structure as in the kidney."

MD cells are positioned to monitor tubular urine NaCl concentration at a point where it is highly variable and where this variability is almost exclusively determined by loop of Henle inflow rate. Because the rate of flow into the loop of Henle is set by GFR and proximal tubular fluid absorption, two variables regulated by extracellular fluid volume, ambient NaCl concentration at the MD reflects body Na⁺ balance, decreasing in volume depletion and increasing in volume expansion.

	FUNCTIONS OF THE JGA

Changes in NaCl concentration in the tubular fluid passing the MD cells have two known effects (Fig. 1). They regulate glomerular vascular tone predominantly in the afferent arterioles and thereby alter glomerular capillary plasma flow and GFR (103). The vascular response is nonlinear and has its greatest sensitivity in the range of ambient NaCl concentrations (13). Because an increase in NaCl concentration or in its determinant, loop of Henle inflow rate, causes a decrease in GFR, the relationship between epithelial and vascular cells is constructed as a negative feedback loop, the tubuloglomerular feedback (TGF) mechanism. The intact or closed TGF loop compensates with near equal efficiency for both increments and decrements in MD NaCl concentration, a consequence of the position of the operating point at the midpoint of the TGF function curve (13, 128).

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Top: relationship between macula densa (MD) Cl concentration and nephron filtration rate. Nephron filtration rates are taken from Ref. 13. MD Cl concentrations are estimated from the relative glomerular filtration rate (GFR) change during retrograde perfusion with varying NaCl concentrations (105). Bottom: relationship between MD Cl concentration and renin secretion in the isolated perfused rabbit juxtaglomerular apparatus (JGA) preparation (data from Ref. 45). GU, Goldblatt unit.

Changes in NaCl concentration at the MD have the second effect of altering the secretion of renin from granular cells: decreases in NaCl concentration cause stimulation of renin release and increases in NaCl concentration inhibit it. Dependence of renin secretion on luminal NaCl concentration, first suggested by Vander (136), has been difficult to prove in intact animals, but its existence has been firmly established by using a perfused JGA preparation isolated from the rabbit kidney, a method that permits an evaluation of the MD influence on renin secretion independent of baroreceptor and adrenergic inputs (120). With the use of this preparation NaCl-dependent renin release rate could for the first time be estimated quantitatively, amounting to ~0.5 nGU · min¹ · meq NaCl¹ (45, 70, 71), where nGU is a nano-Goldblatt unit.

	JGA FUNCTIONS AND NA BALANCE

The coexistence of these two local regulatory pathways, both dependent on the same input, creates a rather complex interrelationship that in some respect may appear maladaptive and teleologically questionable. For example, what could be the advantage of inducing renal vasoconstriction when MD NaCl concentration increases due to volume expansion, and why would this constriction be accompanied by a decrease in renin secretion? Similarly, what would be the advantage of maintaining GFR by renal vasodilation when MD NaCl concentration decreases as a result of blood loss?

The solution to this dilemma is that the TGF system assists in maintaining relative constancy of distal NaCl delivery and NaCl excretion only during fast and random perturbations that are unrelated to body NaCl balance. Once body fluid volume changes have occurred, however, distal NaCl delivery serves to regulate renin secretion and the changed plasma angiotensin II levels support the maintenance of prolonged deviations of distal NaCl concentration from the normal set point. Expressed differently, the two aspects of JGA function are related to Na balance in a temporally sequential manner, with control of afferent arteriolar tone being the major JGA function in the short term and control of renin secretion being its dominant role in the long term (Fig. 2).

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Concept of short-term and long-term function of the JGA. TGF, tubuloglomerular feedback.

Short-Term Function of the JGA: TGF Control of Vascular Tone

View larger version (26K): [in this window] [in a new window]   Fig. 3.   TGF during acute (left) and chronic (right) changes in MD NaCl concentration ([NaCl]MD). Acute changes elicit TGF counterregulation so that GFR and MD NaCl may oscillate around the operating point without changes in mean values. Chronic changes cause TGF adaptation, mainly due to changes in angiotensin II levels resulting from altered renin secretion.

Long-Term Function of the JGA: Control of Renin Secretion

It is a unique and important aspect of MD function that the change in angiotensin II resulting from the altered renin secretion causes TGF regulation to adapt in sensitivity in a way that permits it to serve the needs of body Na⁺ balance (Fig. 3B). This at least in part angiotensin-dependent TGF adaptation prevents the inappropriate responses of reducing steady-state GFR in volume expansion and increasing it in volume depletion. The adaptation in TGF sensitivity also assists in keeping the operating point of the system in the sensitive portion of the feedback curve so that TGF can continue to act as an efficient fast controller of vascular tone although ambient distal NaCl concentration is altered.

The mechanism of MD-mediated renin secretion is suited as a long-term controller: it has a much slower frequency response than the TGF system, more like 1-2 mHz, or about one cycle every 10 min. Furthermore, to the extent that renin acts as a systemic enzyme, changes of renin release rate have to be translated into changes in plasma renin, a process that occurs on the background of the plasma renin pool acting as a buffer. Even when renin secretion is completely interrupted by nephrectomy, considerable amounts of renin are left after prolonged periods of time (8). Persistent changes in MD NaCl concentration cause changes in renin secretion and systemic renin levels that can be sustained for virtually limitless periods of time. For example, members of societies consuming an essentially sodium-free diet maintain increased rates of renin secretion and an elevated plasma renin concentration throughout their lives (81). These long-lasting adjustments in the rate of renin secretion require changes in renin synthesis subsequent to transcriptional or posttranscriptional regulation of renin mRNA expression. Consistent with the role of the renin-angiotensin system as a long-term homeostatic mechanism is the result that changes in renin gene expression and renin mRNA follow a slow time course. After a single injection of furosemide, renal renin mRNA was found to be significantly elevated only after 4 h (17). Similarly, in granular cells in short-term culture, a significant increase in renin mRNA levels after exposure to forskolin was not observed until 4-6 h had elapsed (18, 24). Interestingly, the rise in renin mRNA by forskolin appears to be due to a large extent to an increase in renin mRNA stability (18, 64).

	MEDIATING MECHANISMS IN JGA FUNCTION

The mechanisms mediating between the MD and its target cells, the vascular smooth muscle and granular cells, must be compatible with the characteristics of the two different MD control systems: they must allow rapid alterations in vascular tone in a minute-to-minute fashion, they must explain adaptation of TGF during changes in NaCl balance, and they must be compatible with long-standing tonic effects on renin secretion. In describing the interactive pathways within the JGA, it is useful to distinguish between "mediators" and "modulators" of MD-dependent effects (Fig. 4).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Distinction between mechanisms mediating (top) or modulating (bottom) MD-dependent regulatory events.

A mediator of JGA responses must be able to cause a graded effect on vascular tone or renin secretion, and this gradation must be caused by an NaCl-dependent change in its local concentration, either an increase or a decrease. The local concentration of any modulator, on the other hand, is not primarily dependent on luminal NaCl concentration. It may be generated locally or systemically in a constitutive fashion and may be up- or downregulated by non-MD-dependent mechanisms. The level of local modulator concentrations will determine the characteristics of the effects elicited through the MD-specific mediator pathway.

Response Initiation By NaCl Transport

NaCl transport by MD cells is now relatively well understood, and it resembles in many aspects NaCl transport in thick ascending limb (TAL) cells (Fig. 5). Like in these cells, NaCl uptake is mostly through the Na⁺-K⁺-2Cl cotransporter (NKCC₂ or BSC₁). Conventional electrophysiology and patch clamp evidence has established its presence functionally (66, 98, 99). Presence of the NKCC₂ cotransporter has also been shown at the mRNA expression level using reverse transcription-polymerase chain reaction (RT-PCR) methods (149). The cotransporter expressed in the MD appears to be mainly the B type, the isoform specific for the cortical, but not the medullary, TAL (149). Apical membranes are rich in K⁺ channels, probably of the recently cloned ROMK type, which are Ca²⁺ and pH sensitive (49, 98, 99). There is also a sodium/hydrogen exchanger (presumably NHE₃) in the apical membrane (30). Na⁺-K⁺-adenosinetriphosphatase abundance is less than in TALs, and there are basolateral Cl channels for Cl exit (65, 99).

View larger version (65K): [in this window] [in a new window]   Fig. 5.   Schematic diagram of transport proteins shown to be expressed in MD cells.

Mechanisms of Transport-Related Changes in JGA Function

MD cell composition. When NaCl transport is stimulated by a rise in luminal NaCl concentration, MD cells undergo profound changes in composition and membrane polarity, which include increases in cytosolic Na⁺ and Cl concentrations, increases in cytosolic pH, and membrane depolarization, all of which could potentially be the source for the generation of a transport-dependent downstream signal. However, roles of specific intracellular alterations in the TGF or renin pathway have not been established. In fact, for some, any direct participation seems unlikely. For example, cell alkalinization is unlikely to be causal in eliciting TGF responses, because as mentioned above, TGF responses are not altered by amiloride, although the diuretic prevents the alkalinization normally seen with elevated luminal NaCl concentration. Furthermore, an increase in Cl without an increase in Na⁺ acidifies rather than alkalinizes MD cells (30), but an increased Na⁺ concentration is not required for TGF responses as long as the cation substituting for Na⁺ is small and permeant (105). Finally, preliminary studies in a strain of mice with a null mutation in the NHE₃ gene in which flow-dependent alterations in cytosolic pH cannot occur have not shown significant changes in the TGF response (mice generated by Gary E. Shull, University of Cincinnati, were kindly supplied by John Lorenz).

Interstitial NaCl concentration and osmolality. Because of the absence of blood capillaries and lymphatics in the juxtaglomerular interstitium, changes in local interstitial NaCl concentration may result from changes in NaCl transport and may play a role in mediating MD-dependent responses. On the basis of studies in mesangial cells in culture it has been proposed that interstitial Cl concentration may determine the degree of mesangial cell activation by agonists such as angiotensin II (61, 80). The mesangial cell activation by increased extracellular Cl may result from inhibition of prostaglandin (PG) release, because PGE₂ formation by cultured mesangial cells has been found to be stimulated in a Cl-free medium (80). Furthermore, a decrease in NaCl has been shown to stimulate NO release by mesangial cells, and this inverse relationship between NaCl and NO formation has been suggested to mediate TGF responses (133). However, several arguments could be advanced against both possibilities, the main being that changes in interstitial NaCl concentrations, although probably present in the periglomerular interstitium of amphiuma kidneys, have never been observed in the mammalian JGA (87). Mathematical modeling has shown that juxtaglomerular interstitial NaCl concentration may vary with changes in NaCl transport; however, the determining parameters, transport rate, epithelial water permeability, and interstitial space volume, are not known with the degree of certainty necessary to safely predict interstitial ion concentrations (94). Experimentally, neither luminal nor systemic administration of PGE₂ causes inhibition of TGF responses, as predicted by one of these models (86, 108). Thus it is not clear to what extent mesangial cells in culture are representative of native cells. The finding that mesangial cells release nitric oxide (NO) in a calcium-sensitive manner suggests the presence of either the neuronal or endothelial NO synthase [NOS I or NOS III (133)]. However, in mesangial cells associated with freshly dissected glomeruli, neither NOS I nor NOS III mRNA is found by in situ hybridization or the highly sensitive RT-PCR method (77, 118, 146). Furthermore, NO is unlikely to be a vasodilator mediator of the TGF response because inhibition of NO synthesis has been reported to augment vasoconstriction, not to abolish vasodilation (130, 146).

Changes in interstitial osmolarity could mediate the renin secretory response because swelling and shrinking of juxtaglomerular cells, or probably more specifically, of juxtaglomerular cell granules, has been shown in a number of different preparations to consistently cause changes in renin secretion that are directionally plausible, i.e., a stimulation of renin secretion with decreased tonicity and an inhibition with increased tonicity (35, 119). Experimental evidence has been presented to indicate that changes in cell volume may underlie the renin inhibitory action of agents that cause an increase in cytosolic calcium, such as angiotensin II (62, 63). There is growing support for the concept that the permeability of Ca²⁺-activated Cl and K⁺ channels may provide the critical link between cytosolic calcium and renin secretion. In accordance with this concept, closing of these channels with reduced cytosolic Ca²⁺ would tend to stimulate renin secretion either because of an increase in cell volume or because of some more specific effect of channel closing, such as an increase in cytosolic Cl concentration (58, 62). Conversely, opening of Ca²⁺-activated Cl and K⁺ channels with elevated cytosolic Ca²⁺ would cause cell shrinkage resulting from KCl and water efflux, and this would result in inhibition of renin release (62). The critical question in the context of MD signal mediation is whether external juxtaglomerular osmolarity does in fact vary to the extent necessary, a requirement if NaCl transport-dependent changes in interstitial osmolarity are to represent the direct TGF or renin secretory mediator. It should be pointed out that changes in osmolarity, if they occur, are unlikely to be directly involved in TGF mediation, because the effect of an increase in tonicity on glomerular afferent tone appears to be vasodilation not vasoconstriction (78).

Cellular metabolism. MD-mediated responses may be related to cellular metabolism and the generation of vasoactive metabolites, specifically of adenosine (83). An AMP-specific 5'-nucleotidase has recently been identified that is expressed in MD cells (139). Adenosine deaminase is abundantly present in the glomerulus to clear adenosine from the region and permit the rapid turnoff that is characteristic of the TGF response (85). Adenosine is a constrictor of afferent arterioles. Studies in an isolated perfused vessel preparation have shown that adenosine constricts rabbit afferent arterioles, especially in their distal glomerulus-near region, where the TGF response has been observed to be expressed most clearly (47, 142). Significant vasoconstriction is first seen at a concentration of 10⁸ M, consistent with the concentration dependency of high- affinity adenosine 1 (A₁) receptors. There was no relative relaxation at higher concentrations, suggesting absence of low-affinity adenosine 2 (A₂) receptors in this part of the afferent arteriole. Expression of A₁ receptor, but not A₂ receptor, mRNA has been found in afferent arterioles by in situ hybridization (141). The feasibility of a dependence of adenosine generation on NaCl transport is supported by the finding that the stimulation of Na⁺ transport by glucose in the proximal tubule is associated with a reduction in cellular ATP levels (2) and more specifically by the observation that increased NaCl secretion in shark rectal glands is accompanied by increased adenosine generation and decreased cellular levels of ATP (60).

In addition to this solid background foundation, there is considerable functional evidence in support of a role of adenosine in MD-mediated events, particularly in the TGF response. A₁ receptor blockers such as the xanthine derivative 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) inhibit TGF responses when they are added to the tubular lumen or to the peritubular blood (109). Similar effects have been seen earlier with nonspecific blockers such as theophylline or 3-isobutyl-1-methylxanthine (34, 104). Complete blockade of TGF responses was also observed in an isolated tubule preparation when the nonxanthine A₁ receptor antagonist FK838 was added to either the perfusate or the bath (92). Consistent with the high-level expression of juxtaglomerular A₁ receptors on afferent arterioles is functional evidence that the receptors involved in TGF mediation are probably located outside the tubule. DPCPX was found to fully inhibit TGF responses when it was added to the lumen of neighboring nephrons (109). In this case the blocker must have diffused out of the tubule to affect receptors in the vicinity of the test nephron's glomerulus, probably on its afferent arterioles. Adenosine analogs cause vasoconstriction when added to the lumen (34, 100), and this has been interpreted as indicating an involvement of luminal receptors (34). However, the finding that luminal administration of native adenosine was not associated with augmentation of TGF responses is difficult to reconcile with a role of A₁ receptors in the apical membrane of MD cells in TGF signal transmission (109). Furthermore, it seems implausible to postulate that adenosine is delivered to the MD with the tubular urine in a flow-dependent concentration pattern, mainly because full TGF responses can be elicited without MD cells being exposed to fluid that has previously passed through the proximal tubule and TAL epithelium. Current evidence strongly supports the notion that adenosine is generated by MD cells or TAL cells in their immediate vicinity in a transport-dependent fashion, that adenosine is transported across the basolateral membrane to gain access to the juxtaglomerular interstitium, and that it diffuses through the interstitium to bind to A₁ receptors on afferent arteriolar smooth muscle cells. To the extent that native extraglomerular mesangial cells possess A₁ receptors, it is conceivable that adenosine interacts with mesangial cells and that smooth muscle cells are subsequently activated by gap junctional coupling (50). The increase in cytosolic calcium responsible for afferent arteriolar vasoconstriction is maintained by calcium influx through voltage-activated calcium channels, consistent with the finding that Ca²⁺ channel blockers cause virtually complete inhibition of TGF responses (74). In addition, TGF responses are blunted in the presence of forskolin (5). Figure 6 shows unpublished data from our laboratory demonstrating that forskolin administered by either luminal or peritubular perfusion attenuated, but did not block, TGF responses. Thus a decrease in protein kinase A activity may be an additional component of TGF activation of afferent arteriolar smooth muscle cells.

View larger version (43K): [in this window] [in a new window]   Fig. 6.   Responses of stop flow pressure (PSF) to loop perfusion at 45 nl/min before, during, and after administration of forskolin (104 M). Forskolin was either included in the perfusate (index nephron), added to the peritubular blood (peritubular), or infused into a nephron adjacent to the index nephron (adjacent nephron). Values are means ± SE, and numbers in parentheses indicate number of tubules studied.

Release of ATP by MD cells has been suggested as another coupling mechanism between cell metabolism and TGF responses (79). ATP causes rapid and reversible vasoconstriction of afferent arterioles, more pronounced in juxtamedullary than superficial nephrons, that is mediated through P₂ purinoceptors (53, 142). Involvement of ATP in afferent arteriolar regulation is supported by the observation that P₂ antagonists interfere with autoregulatory adjustments of vascular tone (52). On the other hand, peritubular administration of ATP markedly blunted TGF responses to increased loop perfusion rates (75). It is unclear why the exogenous administration of a potential TGF mediator would eliminate rather than mimic or augment the TGF-induced effect.

Whether adenosine also mediates the acute renin inhibitory response is less clear. The A₁ receptor antagonist DPCPX only partially inhibited the reduction in renin secretion caused by an elevation in luminal NaCl concentration (144). Furthermore, adenosine itself was found to be only a weak inhibitor of MD-stimulated renin secretion in the isolated JGA preparation when added to the bathing fluid (69). Because this effect was augmented by pentastatin, an inhibitor of adenosine deaminase, the marginal magnitude of the adenosine effect may reflect effective degradation of exogenous adenosine by glomerular adenosine deaminase. Whether endogenous adenosine released by MD cells is more shielded against the degradative action of adenosine deaminase is not known.

In summary, adenosine appears to play a central role as a mediator of the acute vascular response to changes in luminal NaCl concentration, but in the renin secretory response it may be somewhat more peripheral (Fig. 7).

View larger version (48K): [in this window] [in a new window]   Fig. 7.   Juxtaglomerular mediators of the TGF response to increased MD NaCl transport (left) and of the renin secretory response to reduced MD NaCl transport (right), and intracellular changes in respective target cells involved in eliciting changes in vascular and secretory endpoints. [Ca]i, intracellular Ca2+; PKA, PKG, protein kinase A and G, respectively; EGM, extraglomerular mesangium.

	ROLE OF NO SYNTHASE

MD cells have been found to be distinctly different in a number of aspects from surrounding TAL cells. Outstanding among these differences is the presence of a constitutive isoform of NO synthase, NOS I, and the expression of the so-called inducible isoform of cyclooxygenase, COX-2 (42, 77, 146). Both NOS and COX products have been suggested to participate in MD-dependent regulation.

NOS I expression in the cortex, as first shown by Wilcox and Mundel and their co-workers (77, 146), is essentially exclusively located in the MD. Presence of NOS I has been demonstrated at the protein level by immunofluorescence, at the mRNA level by in situ hybridization and RT-PCR, and at the enzyme activity level by the NADPH-dependent diaphorase reaction (11, 77, 118, 132, 146). Diaphorase positivity of MD cells had been noticed earlier without being interpreted as representing NOS activity (15).

NO and the TGF Mechanism

The availability of mice with a null mutation in the NOS I gene provided an opportunity of studying the effect of chronic and selective NOS I deficiency in regulating TGF responses (48). Experiments from our laboratory indicate that NOS I-deficient mice have a steady-state TGF response of stop flow pressure that is not markedly different from that of control mice (Fig. 8). Thus deficiency in MD-generated NO formation does not appear to alter TGF responses dramatically under these very chronic conditions. It is possible that some secondary alteration provides a compensatory input that counteracts the TGF-enhancing effect of NO deficiency. For example, we and others have noted that NOS I-deficient mice have a markedly reduced level of renal renin content and renin mRNA expression so that a reduction in angiotensin-dependent TGF responsiveness may have nullified the TGF augmenting action of NO absence (41). It is also possible that another NOS isoform is upregulated in the NOS I knockout animals, a possibility that has not been explored so far. Finally, a residual, enzymatically active NOS I protein has been detected in the brain of NOS I knockout mice at a low level (14). This protein has been shown to be translated from a NOS I mRNA splice variant that lacks exon 2, the part of the NOS I gene targeted in the gene-disruption approach (48). Our observations would be inconclusive if this NOS I isoform is in fact present in MD cells of the knockout mice.

View larger version (39K): [in this window] [in a new window]   Fig. 8.   Responses of PSF (PSF) to graded increases in loop perfusion rates in control mice [nitric oxide synthase (NOS) I +/+] and in mice with a null mutation in the NOS I gene (NOS I /) (unpublished data from our laboratory).

Unless truly specific NOS blockers are available, difficulties in interpretation arise from the fact that NOS enzymes are not only expressed in MD cells, but also in other parts of the JGA, most notably in endothelial cells. Thus the source of any NO involved in the TGF response cannot be identified with certainty. Another important issue that needs to be resolved is whether NO is generated in the juxtaglomerular region in a NaCl-dependent fashion, increasing or decreasing with increased or decreased NaCl concentration. In the absence of direct measurements of NO in the JGA, the currently available indirect evidence does not favor the hypothesis that release of NO from MD cells varies with changes in luminal NaCl concentration or NaCl transport rate. Cytosolic Ca²⁺, the obligatory physiological activator of constitutive NOS isoforms, was not found to be consistently altered by luminal NaCl concentration or transport in the MD (7, 95). Furthermore, previous studies employing blockers of calmodulin have failed to demonstrate an augmentation of TGF responses, although this would have been the predicted outcome if full TGF responsiveness is tonically suppressed via a Ca²⁺-calmodulin-dependent activation of NOS (6). On the other hand, increased release of NO from endothelial cells during TGF activation is conceivable, because afferent arteriolar constriction is predicted to increase local shear stress, a factor known to stimulate NO formation.

The regulation of local NO production must not necessarily occur at the level of NOS activity. It has been proposed that the uptake or availability of arginine may become rate limiting in the formation and release of NO (26). Furthermore, an exciting new area of NO regulation has been opened by the discovery of several endogenous inhibitors of NO synthases, such as the methylated arginine derivatives N^G,N^G-dimethly-L-arginine (ADMA) or N^G-nitro-L-arginine methyl ester (L-NAME), and more recently of protein inhibitor of neuronal NOS (PIN), an endogenous NOS inhibitor that may be specific for NOS I (56, 134). It is obvious that regulation of NOS activity by endogenous inhibitors could be critical for local NO formation, but little information is currently available to indicate whether ADMA or L-NAME is present in the JGA and how it is regulated. PIN mRNA has been found in our laboratory to be widely expressed in the kidney, including in the MD segment, but its major expression site appears to be the proximal tubule (unpublished data from our laboratory).

NO and Renin Secretion

Isolated juxtaglomerular cells in short-term culture respond to brief exposure of nitroprusside with a dose-dependent inhibition of renin secretion that we interpret as the result of a direct inhibition of renin release by NO (38, 114). It is likely that the inhibitory effect of NO on renin release is mediated by guanosine 3',5'-cyclic monophosphate (cGMP) resulting from activation of the NO target guanylate cyclase. Sodium nitroprusside (SNP) causes a prompt rise in cGMP levels in cultured juxtaglomerular cells, and activation of cGMP-dependent kinase inhibits renin secretion through mechanisms that are presently not known in detail (38, 114). In contrast, when renin secretion was measured in an isolated perfused JGA preparation during perfusion with a low NaCl concentration, the luminal addition of L-arginine stimulated renin secretion and this stimulation was abolished by NOS blockade, suggesting that in this setting NO is renin stimulatory (44). Consistent with this conclusion is our observation that the NaCl dependency of renin secretion was essentially abolished in the presence of an NOS blocker in the tubular lumen, a change that was due entirely to prevention of the rise of renin secretion caused by a low luminal NaCl (44). It appears therefore that a stimulatory input of NO, possibly released from MD cells, is a requirement for the stimulation of renin secretion by low luminal NaCl or low NaCl transport rates. This finding is compatible with observations showing that the stimulation of renin secretion by furosemide as well as the stimulation of renin secretion by a low-salt diet, both presumably MD-mediated events, could be inhibited by the administration of NOS inhibitors including the somewhat NOS I-specific 7-nitroindazole (3, 91, 115). Furthermore, the increase in renin release from renal microvessels isolated from rats treated with furosemide was found to be prevented by NOS inhibition (16). The conclusion that a low NaCl concentration at the MD stimulates renin secretion in an NO-dependent fashion is supported by findings showing that the increased renin secretion caused by a reduction in arterial or perfusion pressure in kidneys of conscious dogs and in isolated rat kidneys was markedly and consistently blunted by NOS inhibition (88, 110). One would expect MD NaCl to fall at low arterial pressures and this could contribute to the stimulation of renin secretion. Furthermore, the presence of an NOS inhibitor prevented the stimulation of renin secretion observed in isolated perfused rabbit arterioles during low perfusion pressures (Fig. 9).

View larger version (22K): [in this window] [in a new window]   Fig. 9.   Effect of NG-nitro-L-arginine (NNA), a nonspecific NOS inhibitor, on the relationship between perfusion pressure and renin secretion in isolated perfused arterioles from rabbits. Data included in this graph represent the subset of arterioles (~50%) in which an increase in perfusion pressure reduced renin secretion (unpublished data from our laboratory generated by X.-R. He).

The mechanism of the stimulatory effect of NO on renin secretion is currently under intense investigation. It is possible that NO released from the MD interacts with cells other than granular cells and that this interaction results in the formation of a renin secretagogue, for example of a PG, so that the stimulatory effect of NO would be indirect. However, earlier studies have shown that exposure of cultured granular cells to SNP for longer than 4 h results in stimulation of renin secretion, suggesting that NO can also be a direct activator of renin secretion (114). Because cellular levels of cGMP were found to be elevated compared with control, a change that in itself should inhibit renin release (25), one must assume that long-term SNP exposure is associated with either a decrease in granular cell Ca²⁺ or an activation of granular cell adenosine 3',5'-cyclic monophosphate (cAMP)-dependent kinase, most likely a combination of both. NO has been shown to be capable of reducing cytosolic calcium through different mechanisms (9, 23, 36, 116), and the importance of the reduced cytosolic calcium is probably to prevent opening of Ca²⁺-activated K⁺ channels due to lowering of the calcium activation threshold caused by NO via protein kinase G (124). The pathway of the postulated activation of protein kinase A is unclear. Elevated cGMP levels may reduce cAMP degradation by affecting PDE III, a cGMP inhibitable isoform of phosphodiesterase (20, 43). However, because intracellular levels of cAMP were not markedly elevated by chronic treatment of granular cells with SNP (114), stimulation of the kinase A pathway may result from crossreaction with cGMP or with some other unknown activator. There is evidence to suggest that protein kinase A may be activated by cGMP in rat aortic smooth muscle cells and in an intestinal cell line (22, 29). The fact that forskolin was found to be unable to stimulate renin release from isolated renal microvessels of L-NAME-treated rats supports the notion that NO stimulation of renin release occurs mainly through the protein kinase A pathway (16). The importance of this pathway in MD-stimulated renin release is also highlighted by our observation that maintenance of high cellular levels of cAMP by forskolin completely abolished the inhibitory effects of increasing MD NaCl in the isolated JGA preparation (Fig. 10; H. Weihprecht, J. Lorenz, J. P. Briggs, and J. Schnermann, unpublished observations).

View larger version (30K): [in this window] [in a new window]   Fig. 10.   Effect of increased luminal NaCl concentration on renin secretion in the isolated perfused rabbit JGA preparation in the absence (top) and presence (bottom) of 105 M forskolin in the bathing medium (unpublished data from our laboratory generated by H. Weihprecht).

In summary, juxtaglomerular NO formation appears to be a critical determinant of the stimulation of renin secretion by low-MD NaCl transport. Although the universal receptor for NO is soluble guanylate cyclase, a simultaneous stimulation of the protein kinase A pathway and a reduction of cytosolic calcium in granular cells appear to be the required intracellular mediators of the NO effect on renin secretion (Fig. 7).

	ROLE OF COX

The local production of PGs has been implicated in MD responses for some time, but the recent demonstration of a specific expression of COX-2 in MD cells has sparked a renewed interest in this topic (42). Assessing mRNA levels by in situ hybridization using radiolabeled riboprobes, Harris and co-workers (42) reported that COX-2 mRNA, the isoform typically only found after induction by inflammatory cytokines, was expressed in a very localized fashion in MD cells under basal conditions. This finding has been confirmed by Guan et al. (40) in the rabbit and by Dr. Arend (unpublished) in our laboratory with digoxigenin-labeled riboprobes, as well as with RT-PCR methods. In fact, MD cells and some TAL cells in their vicinity (40, 42, 138) are the only cells in the cortex expressing COX-2 under basal conditions, so that basal COX-2 levels in the renal cortex as a whole are actually quite low compared with the inner medulla, the main region of renal COX-2 expression (150). Renal cortical COX-1 mRNA, on the other hand, is only found in glomeruli, presumably in mesangial cells (150).

In situ hybridization studies of PG receptor mRNA expression has revealed the presence of PGE₂ receptors of the EP₂ subtype in the glomerulus and of the EP₃ subtype in the MD (126). As assessed by RT-PCR, microdissected glomeruli also express the mRNA for PGF₂ and thromboxane receptors (1; unpublished data from our laboratory).

PGs and the TGF Mechanism

Recent evidence suggests that 20-hydroxyeicosatetraenoic acid (20-HETE), a cytochrome P-450 product of arachidonic acid, may be another possible acute mediator of the TGF response (152). Two inhibitors of cytochrome P-450 enzymes, 17-octadecynoic acid (ODYA) and clotrimazole, caused a marked attenuation of the TGF response when added to the tubular lumen for an extended period of time (152). To act as a mediator in the TGF pathway one would have to postulate that 20-HETE is delivered to the afferent arteriolar smooth cells in an MD NaCl transport-dependent fashion. No direct evidence is available that would speak to this question. Renal epithelial formation of 20-HETE has been shown to be almost exclusively a function of the proximal tubule, and this localization is fully consistent with our observation that renal mRNA expression of CYP4A₁ and CYP4A₂, cytochrome P-450s thought to be responsible for 20-HETE production in the kidney, was localized predominantly in the proximal tubule and in the loops of Henle (82; unpublished data from our laboratory). Importantly, 20-HETE formation as well as CYP4A₁ and CYP4A₂ expression was not found in glomeruli, making the local production by MD cells or other cells in the JGA questionable (82; unpublished data from our laboratory). It is possible that delivery of 20-HETE to the JGA occurs through the tubular lumen, but it may be pertinent to reiterate that TGF responses do not require the presence of a perfused loop for the delivery of a potential mediator, as clearly shown by the retrograde perfusion studies and by the in vitro observations from Ito's group (55). 20-HETE has been shown to inhibit TAL NaCl transport by inhibiting Na⁺-K⁺-2Cl cotransport and by closing K⁺ channels (27, 140, 151). If such an inhibition of NaCl transport also occurred in MD cells, any increase in 20-HETE production would not be expected to enhance, but to attenuate, TGF responses. Thus it seems possible that ODYA and clotrimazole act by causing a reduction in afferent arteriolar tone, representing another example for TGF inhibition by nonspecific vasodilation (103).

PGs and Renin Secretion

4

As with the origin of NO, the presence of the two COX isoforms in different cells of the JGA makes it difficult to pinpoint the origin as well as the target cells of locally produced PGs. Recent evidence in mice showing that the specific COX-2 blocker NS-398 is able to essentially completely inhibit the stimulation of renal renin content by a low-salt diet (41) appears to identify COX-2 as the isoform involved and, by implication, MD cells as the source of the released PGs because COX-2 does not appear to be expressed in native mesangial cells (40, 42, 138, 150).

If, as current evidence suggests, MD-stimulated renin release results from the action of both locally produced NO and PGs, the question of how the action of these two mediators is coordinated needs to be resolved. A substantial body of experimental evidence indicates that NO can be an activator of COX-2 in a number of different cells and organs (21, 31, 51, 93, 96, 97), and it is possible therefore that NO and PGs are arranged in series, with NO from MD cells modifying the generation of PGs by, for example, mesangial cells. Whether acting in parallel or in series, NO and PGs appear to be the major determinants in the acute response of renin secretion to a change in MD NaCl.

	TGF ADAPTATION

As has been pointed out earlier, an aspect of TGF control that is vital in understanding its integration into the mechanisms responsible for body fluid homeostasis is its adaptation to more long-standing perturbations of NaCl concentration at the MD. The typical physiological challenge of this type results from alterations in body Na⁺ content, with volume expansion causing increased, and volume depletion causing decreased, MD NaCl concentrations, changes that initially are the result of decreased or increased absorption of NaCl in proximal tubules and loops of Henle. Such more long-standing changes in MD NaCl cause the TGF mechanism to alter its characteristics: during chronically reduced MD NaCl the system responds to lower flows with larger changes of SNGFR, and during chronically increased MD NaCl it responds to higher flows with smaller changes of SNGFR (103). Formally, two types of adaptations in the TGF system can be distinguished. Resetting refers to shifts in the position of the TGF function (the relationship between SNGFR and NaCl at the MD) or in the range of MD NaCl concentrations over which the system is operating, whereas sensitivity changes refer to altered responsiveness as expressed by altered slope and/or maximum response magnitude. Alterations in salt balance appear to cause both resetting and sensitization/desensitization of the TGF system. The time frame of the adaptive response to increments in MD NaCl concentration has been recently studied in single nephrons perfused at elevated flow rates for extended periods of time (129). In these studies, resetting developed over the initial 30-40 min of hyperperfusion, whereas desensitization occurred in the time between 40 and 60 min. Similar dynamics of TGF adaptation have been reported earlier in studies where acute volume depletion of chronically volume-expanded animals restored normal TGF responses within 60-120 min (76). The importance of TGF adaptation for body fluid balance has been difficult to study experimentally. Nevertheless, failure of the TGF system to undergo appropriate adaptation during volume expansion has been implicated in the development of arterial hypertension in Milan hypertensive rats (10).

Several mechanisms have been proposed to be responsible for TGF adaptation, but the most important single factor is probably the level of renin and angiotensin II. Angiotensin receptor blockers as well as inhibitors of converting enzyme in relatively high doses blunt the magnitude of the TGF response by ~50-60% (89, 125). The infusion of angiotensin II, on the other hand, either intravenously or peritubularly, enhanced TGF responsiveness, a property not shared by other vasoconstrictors, such as vasopressin or norepinephrine (73, 101). Furthermore, when the TGF response was suppressed by an acute volume expansion, the administration of angiotensin II restored TGF responses in a dose-dependent manner (102). Overall, TGF responses expressed as percent reduction of stop flow pressure can be shown to correlate with increasing plasma angiotensin II levels and maybe local tissue levels of the peptide (102). This important modifying role has recently been corroborated in our laboratory by studying TGF responses in mice with a null mutation in the AT_1A receptor, the major renal receptor for angiotensin II (54). In wild-type mice, maximum TGF responses were ~9 mmHg, with the late proximal flow rate associated with half-maximum response being ~8 nl/min, a flow rate corresponding to the normal late proximal flow rate in the mouse nephron (107). In contrast, TGF responses in homozygous AT_1A receptor knockout mice were essentially completely abolished, whereas mice heterozygous for the gene deletion had intermediate responses (107). Thus it appears that an AT_1A receptor-mediated effect of angiotensin II is a required constituent of the TGF pathway also under chronic conditions. Although it is possible that other factors contribute to TGF adaptation, angiotensin II is likely to play a dominant role because its generation is directly coupled to the changing NaCl signal and therefore provides a quasi-automatic link between MD NaCl and TGF sensitivity.

The mechanism of the specific TGF-modifying role of angiotensin II is not clear, but one possibility is that it is an expression of the specific synergistic interrelationship between the actions of angiotensin II and the TGF mediator adenosine. Renin-angiotensin system-dependent modulation of the actions of adenosine in the in vivo kidney has been demonstrated in several studies in which the constrictor effect of adenosine was found to be attenuated or blocked in salt-loaded animals and in animals treated with angiotensin receptor antagonists (84, 121). Furthermore, the vasoconstriction caused by luminal application of the A₁ agonist cyclohexyladenosine was greatly blunted in AT_1A receptor knockout mice (unpublished data from our laboratory). In vitro evidence supports the reality of this interaction. For instance, the constrictor effect of angiotensin II in isolated afferent arterioles from the rabbit was found to be reduced by the addition of the A₁ receptor antagonist DPCPX (143). In addition, exposure of afferent arterioles to submaximal concentrations of angiotensin II and adenosine causes vasoconstriction that is synergistic in nature, that is, the effect of the two agents combined is greater than additive (143).

In summary, in response to relatively high-frequency perturbations, there are short-term fluctuations of vascular resistance resulting from oscillations of MD NaCl around the operating point, with renin release being stable over time. These changes in resistance are primarily adenosine dependent. However, when MD NaCl is displaced for a longer time period, for example by changes in salt balance, the sensitivity of the response of afferent resistance changes, and this adaptive response is at least to a large extent due to the other consequence of an altered MD NaCl, the change in renin secretion.

	MD CONTROL OF GENE EXPRESSION

It has become evident in recent years that the distinct cytology of MD cells is accompanied by the presence of a similarly distinct pattern of proteins reflecting a highly cell-specific gene regulation. Compared with neighboring cells, MD cells are unique in expressing NOS I, COX-2, oxytocin receptors, parathyroid-hormone related peptide, and integrin 6 and lacking in the expression of Tamm-Horsfall protein and epidermal growth factor (103). It is conceivable that this cell-specific expression pattern is another result of luminal fluid/MD cell interactions. This possibility is particularly attractive for those products that change in expression with chronic changes in the level of MD NaCl concentration. NaCl-dependent regulation of gene expression in MD cells may be the starting point for regulation of genes in more distal juxtaglomerular cells, such as renin in granular cells, or for control of juxtaglomerular cell growth and recruitment (Fig. 11).

View larger version (27K): [in this window] [in a new window]   Fig. 11.   Concept of MD-regulated gene expression: prolonged alterations in MD NaCl concentration caused, for example, by changes in NaCl intake, renal artery constriction, diuretics, or ureteral obstruction, direct alterations in the mRNA expression of genes with high representation in the JGA such as cyclooxygenase-2 (COX-2), NOS I, or renin.

NOS I

Kidney development is another situation where local factors may by involved in MD-specific expression of NOS I. NOS I mRNA was only weakly expressed in nephrons before the glomerular capillary loop stage and greatly increased in the MD of more mature nephrons, suggesting that nephron flow may be a prerequisite for MD expression of NOS I (28).

COX-2

Renin

Upregulation of renin gene expression by NO was first suggested by studies in isolated granular cells where the NO donor SNP caused an increase in renin mRNA levels (24). Tonic in vivo stimulation of renin expression by NO derived from the MD is suggested by the recent observations that renal renin content is reduced in NOS I-deficient mice (41). Furthermore, NOS inhibition with L-NAME has been found to completely block the increase in renin mRNA after administration of furosemide for 4 days by minipump infusion (115). Similarly, the increase in renin content in renal microvessels caused by a 5-day furosemide treatment was completely prevented by L-NAME (127). Finally, L-NAME reduced the increase in renin mRNA in the clipped kidneys of Goldblatt hypertensive rats (111). NOS inhibition prevents the increase in renal renin content and mRNA caused by the converting enzyme blocker ramipril, indicating an important contribution of NO to the renin stimulatory action of angiotensin-converting enzyme blockade (113, 127).

Experimental tools to study the potential role of COX-2 products have only recently become available. Consequently, less is known about the role of COX-2 products in regulating renin gene expression. Nevertheless, the specific COX-2 inhibitor NS-398 has been reported to prevent the increase in renal renin content caused by a low dietary Na⁺ intake (41). Anti-inflammatory nonsteroidal drugs prevent the full effect of furosemide administration and renal artery clipping on renal mRNA expression (112, 115).

The molecular mechanisms responsible for the coordinated renal juxtaglomerular induction of NOS I, COX-2, renin, and probably other genes under conditions of decreased total body Na⁺ are not known at present, although it is likely that the cAMP-protein kinase A pathway is involved in renin gene transcription or renin mRNA metabolism. It will be an interesting challenge to determine the interrelationships in the regulation of these juxtaglomerular marker genes and the importance of expression changes for MD-dependent events.

Perspectives

Heterogeneity of structure and function has long been recognized as a characteristic feature of the organization of the kidney. Whereas it has become clear that differences in cellular protein representation are responsible for this heterogeneity, the factors controlling the cell specificity of expression are largely unknown. A typical example is the MD plaque where a small number of cells in each nephron is highly distinct from their neighboring cells. Furthermore, in response to variations in salt intake these cells adjust their expression pattern in a way that can be quite different from the type of change found in other cells of the kidney. It seems reasonable to suspect that changes in luminal NaCl concentration not only cause marked alterations in JGA function, but that these changes, if they persist, also direct gene expression in MD and other juxtaglomerular cells. The addition of this molecular dimension to the variables controlled by luminal fluid composition is expected to be necessary to fully understand the role of the JGA in salt and water homeostasis.