Adenosine formed by 5′-nucleotidase mediates tubuloglomerular feedback

Nephron function is stabilized by tubuloglomerular feedback (TGF). TGF operates within the juxtaglomerular apparatus, sensing changes in tubular flow and eliciting compensatory changes in single nephron GFR (SNGFR). The mediator(s) of TGF remains unconfirmed. One theory is that ATP consumed in active transport by the macula densa leads to formation of adenosine, which causes glomerular vasoconstriction. We performed micropuncture in rats to test this hypothesis. Adenosine activity was manipulated by microperfusing nephrons with adenosine A1 receptor blocker, A1-agonist, or 5′-nucleotidase inhibitor. Effects on TGF were characterized by changes in TGF efficiency (the compensation for small perturbations in tubular flow) and by changes in the maximum range over which TGF can cause SNGFR to change. These data were further applied to generate TGF profiles [SNGFR versus late proximal flow (V_LP)]. TGF efficiency was significantly reduced by blocking A1-receptors. TGF efficiency, TGF range, and the slope of the TGF profile (ΔSNGFR/ΔV_LP) were all significantly reduced by blocking 5′-nucleotidase. When adenosine activity was clamped by combining 5′-nucleotidase inhibitor with A1-agonist to determine whether TGF requires adenosine to be present or to fluctuate, the TGF slope was reduced by 83%, indicating that adenosine activity must fluctuate for normal TGF to occur and that adenosine is a mediator of TGF.

Glomerular filtration is coupled to tubular reabsorption by a system of tubuloglomerular feedback (TGF). TGF operates within the juxtaglomerular apparatus (JGA) of each nephron, where changes in the salt content of tubular fluid at the end of Henle’s loop are sensed and transmitted to the glomerular microvasculature to evoke compensatory changes in single nephron GFR (SNGFR). The molecular mediator(s) of TGF has not been confirmed, but one of the main theories to account for TGF is the so-called adenosine hypothesis. According to the adenosine hypothesis, the consumption of ATP during NaCl transport across the macula densa causes adenosine to accumulate in the region of the macula densa, from where it diffuses to the vascular pole of the glomerulus and causes vasoconstriction of the preglomerular arteriole (1, 2). This hypothesis has teleologic appeal because linking SNGFR to the ATP/ADP ratio in the tubule will protect the tubule against accruing negative energy balance regardless of whether energy balance is threatened by an increase in SNGFR or by a decrease in the ability of the tubule to generate ATP. In organs other than the kidney (especially the coronary circulation), adenosine is well established as an endogenous vasodilator that ensures blood flow to match metabolic demand. The adenosine hypothesis of TGF postulates a role for adenosine that is analogous to other organs in the sense of providing a link between blood flow and metabolism. However, the adenosine hypothesis of TGF requires adenosine to cause renal vasoconstriction, not vasodilation. This is because blood flow in the kidney is a main determinant of metabolic work.

There are published data consistent with the adenosine hypothesis of TGF. For example, activation of adenosine A1 receptors with exogenous agonists causes vasoconstriction of the preglomerular arterioles (3) and postischemic renal vasoconstriction can be blocked with A1-receptor antagonists (4). Furthermore, endogenous adenosine must be present for TGF to function normally, as A1 receptors blockers can prevent the TGF–mediated decline in glomerular capillary pressure that normally occurs when the loop of Henle is perfused at supraphysiologic flow rates with artificial tubular fluid (2). However, the fact that adenosine must be present in order to elicit the maximum TGF–mediated change in glomerular capillary pressure need not imply that adenosine mediates TGF. For example, it is possible for TGF to affect SNGFR independent of glomerular capillary pressure (5, 6), and there are other vasoconstrictors, most notably angiotensin II, that enhance the TGF–mediated change in glomerular capillary pressure but do not mediate the TGF response (6–8).

The present studies were performed to test the adenosine hypothesis more definitively. A variety of in vivo micropuncture approaches were used to measure the incremental TGF response to small perturbations in ambient tubular flow and to assess the maximum range over which SNGFR can be made to change by manipulating TGF. To exclude the possibility that adenosine is a mere background requirement of the TGF system, drugs were administered at the single nephron level in various combinations to block adenosine A1 receptors, prevent adenosine formation by 5′-nucleotidase, or clamp adenosine activity. The results of these experiments confirm that adenosine is a mediator of TGF.

Data were obtained from 50 rats weighing 255–346 g. Mean arterial blood pressure was 110 ± 3 mmHg.

Experiments in free-flowing nephrons. Paired perturbation cycles (control ATF followed by ATF + drug) were completed in 43 free-flowing nephrons from 17 rats (Figures 1 and 2). Of these, eight nephrons were tested for effects of CHA, ten nephrons were tested for effects of KW3902, 13 nephrons were tested for effects of MADP, and 12 nephrons were tested for effects of MADP + CHA (ADO clamp). Baseline V_LP before administration of drug(s) was 18.1 ± 1.1 nL/min and was not different between groups (P = 0.7 for the effect of group on baseline V_LP by ANOVA). Baseline TGF compensation among the 43 nephrons was 0.713 ± 0.061, with some variability between groups mostly attributable to high baseline compensation among nephrons in the KW3902 group.

Figure 1

Ambient tubular flow before and during addition of various drugs to the free-flowing late proximal nephrons (mean ± SEM). Fluid was drawn from the tubule directly upstream from the pump delivering the drug(s) to correct for the volume required to deliver the drug(s). P values refer to the effect of each drug by repeated measures ANOVA. CHA, adenosine A1 receptor agonist (cyclohexyladenosine); KW3902, adenosine A1 receptor blocker; MADP, inhibitor of 5′-nucleotidase (α, β methylene adenosine diphosphate); ADO clamp, MADP + CHA.

Figure 2

Fractional compensation for ± 5 nL/min perturbations in V_LP applied before and during addition of various drugs to the free-flowing late proximal nephrons (mean ± SEM). P values refer to the effect of each drug by repeated measures ANOVA. The adenosine A1 agonist, CHA, had no effect. The blocking adenosine A1 receptors with KW3902 or inhibiting 5′-nucleotidase with MADP clearly reduced fractional compensation. The effect of the ADO clamp was equivocal by repeated measures, although fractional compensation during the ADO clamp was significantly less than a control value pooled from all groups. ^AP = 0.02 by unpaired Student’s t test versus homogeneous pooled controls.

The effects of manipulating the adenosine system on V_LP and fractional compensation (C) were tested by repeated measures ANOVA. During perfusion with CHA, V_LP declined by 29 ± 7% (P = 0.003), whereas C was unaffected (P = 0.5). During perfusion with, KW3902, V_LP increased by 17 ± 5% (P = 0.02) and C declined by 44 ± 18% (P = 0.013). During perfusion with MADP, ambient V_LP increased by 30 ± 8% (P = 0.001) and C declined by 24 ± 6% (P = 0.005).

MADP + CHA were perfused together with the intent of maintaining ambient tubular flow constant. However, during these perfusions, ambient V_LP declined by 27 ± 6% (P = 0.0005), indicating that the perfusate contained too much CHA to achieve an adenosine clamp at the original level of activity. This maneuver tended to diminish fractional compensation, although the statistical significance was less clear-cut than for MADP alone or for KW3902.

Within each group, and across all groups, there was a significant correlation of compensation with V_LP. (When VMFV data are obtained over a range of V_H and fit with a polynomial, peak compensation and V_LP are uncorrelated [4]. When compensation is calculated by linear regression based on a finite perturbation, this will underestimate the true maximum value, and this underestimation will be exaggerated at lower V_LP. Hence, the C and V_LP appear correlated in the present case.) Correcting for this confounding effect of V_LP revealed a tendency for the ADO clamp to reduce the fractional compensation (P = 0.098). Also, as discussed earlier here, there was intergroup heterogeneity for predrug compensation that disappeared when the A1-blocker group was excluded from the analysis. Pooling the remaining controls and comparing these to the clamp by unpaired t test confirmed that compensation was reduced during adenosine clamp (P = 0.02).

Orthograde perfusion in wax-blocked nephrons. SNGFR and V_LP were measured during minimum and maximum TGF stimulation in multiple nephrons from 17 rats (Figures 3–5). Collections were made during perfusion at 38 and 4 nL/min in 84 nephrons. Data from six nephrons were excluded because tubular fluid to plasma inulin ratios were not consistent with late proximal collections. The remaining 78 nephrons included 19 perfused with MADP, 19 perfused with MADP + CHA, and 40 controls. These experiments were performed after it was known that the 10–50 μM CHA used in the free-flow clamp studies reduced ambient tubular flow. For these studies, the clamp was performed with 1 mM MADP and 2 μM CHA.

Figure 3

SNGFR during loop of Henle perfusion at 4 nL/min (SNGFRmax) versus flow during loop perfusion at 38 nL/min (SNGFRmin); mean ± SEM. Stimulation of TGF through loop of Henle perfusion caused SNGFR to decline in all cases.

Figure 5

Glomerular tubular balance. The TGF system was used as a tool to manipulate SNGFR. Values are shown corresponding to paired determinations of SNGFR and V_LP or absolute proximal reabsorption (APR) during minimal and maximal stimulation of TGF. Controls are indicated by circles; MADP, by squares; ADO clamp, by triangles. The right-hand panel depicts the slopes of the lines in the left-hand panel that correspond to f₂′ in Eq. 2 (see text). P values shown refer to comparisons with control by Student’s t test with Bonferroni correction for multiple group comparisons.

Maximum TGF responses. To test for adenosine as a determinant of the maximum TGF response, intergroup comparisons were performed for both absolute changes in SNGFR (ΔSNGFR) and normalized changes in SNGFR (ΔSNGFR/SNGFRmean). For perfusion with ATF, SNGFRmean was 23.8 ± 1.6 nL/min, ΔSNGFR was 14.5 ± 1.3 nL/min, and ΔSNGFR/SNGFRmean was 0.70 ± 0.06. In nephrons perfused with the 5′-nucleotidase inhibitor, MADP, SNGFRmean was 30% greater than control (P = 0.045 by ANOVA with Tukey test for multiple group comparisons). At the same time, ΔSNGFR was less by 36% (P = 0.089) and ΔSNGFR/SNGFRmean was less by 48% (P = 0.01) than in control nephrons. In these ADO-clamp experiments, the goal was achieved of not affecting SNGFRmean (P = 0.99 versus control). However, although SNGFRmean remained unchanged, the ADO clamp reduced ΔSNGFR by 52% (P = 0.008) and reduced ΔSNGFR/SNGFRmean by 52% (P = 0.004). These results are depicted in Figures 3 and 4.

Figure 4

Effects of MADP and ADO clamp on SNGFR at the TGF inflection point (SNGFRmean), the maximum amount by which SNGFR can be made to change via activation of TGF (ΔSNGFR), and ΔSNGFR normalized to SNGFRmean (mean ± SEM). P values shown are comparisons to control by ANOVA with Tukey adjustment for post hoc intergroup comparisons.

Analysis of GTB. Values for ΔAPR and ΔV_LP were calculated for each nephron along with values for ΔSNGFR. The relationship between these parameters and ΔSNGFR were quantified for the nephrons in each group. As expected, both ΔAPR and ΔV_LP correlated strongly with ΔSNGFR. The slopes of the GTB curves (ΔV_LP/ΔSNGFR) were 0.60 ± 0.05, 0.88 ± 0.11, and 0.99 ± 0.13 in control, MADP-perfused, and clamped nephrons, respectively. After Bonferroni correction for multiple group comparisons, MADP and ADO-clamped nephrons differed from control at P = 0.052 and P < 0.02, respectively. These results are depicted in Figure 5.

Loop of Henle reabsorption. Nephrons were perfused from the late proximal tubule at 16 nL/min with ATF followed by ATF + CHA (n = 10), ATF + MADP (n = 10), or ATF (time controls, n = 9). The time-control experiments confirmed a stable preparation and drug perfusion experiments revealed no effect of CHA or MADP on reabsorption of water, sodium, chloride, or potassium from the loop of Henle (Table 1).

Table 1

Content of early distal fluid during orthograde perfusion of Henle’s loop at 16 nL/min with artificial tubular fluid

Effects on the TGF slope. From the fractional compensation and GTB slopes, the mean ± SEM operating-point slope of the TGF curve was calculated for each group to yield slopes of 4.2 ± 1.2, 1.0 ± 0.3, and 0.7 ± 0.2 for control, MADP, and ADO clamp, respectively. After Bonferroni correction for multiple group comparisons, both MADP and ADO clamp differed significantly from control (P ≤ 0.02 for each comparison). In other words, the steepness of the slope of the TGF curve was markedly reduced by blocking 5′-nucleotidase, and there was no tendency for the TGF slope to be restored when enough adenosine agonist was added back to reverse the effect of MADP on SNGFRmean. TGF curves derived by matching ΔSNGFR and ΔV_LP/ΔSNGFR from the orthograde perfusions with fractional compensation measured in free-flowing nephrons are depicted in Figure 6. To generate these curves, it was assumed that the fractional compensation reflected the behavior of TGF at the inflection point.

Figure 6

TGF curves generated from the present data. Four parameters are required to generate each curve. These include SNGFRmean, ΔSNGFR, V_LP at the inflection point, and f₁′ at the inflection point. To generate these curves, it was assumed that f₁′ is a symmetric sigmoid and that nephrons in the perturbation experiments were operating near the inflection points of their respective TGF curves. All other features of the curves are obtained directly from the present data.

Retrograde perfusion experiments. SNGFR was measured in each nephron during perfusion with low-salt distal ATF (minimum TGF) and high-salt distal ATF (maximum TGF) in 32 nephrons from six rats (15 control nephrons, 17 ADO-clamp nephrons) (Figure 7). The ADO-clamp perfusate contained 0.5 μM CHA + 1 mM MADP in five nephrons and 1.0 μM CHA + 1 mM MADP in 12 nephrons. Values for SNGFRmean were calculated for each nephron analogous to those calculated for the orthograde perfusion studies (see earlier discussion here). There was a slight tendency for the clamp solution with the higher CHA content to manifest a lower SNGFRmean (21 ± 1 vs. 23 ± 2 nL/min; P = NS). Overall, the ADO clamp achieved the intent of not altering SNGFRmean (22 ± 1 vs. 23 ± 2 nL/min, clamp versus control, P = NS.) However, ADO clamp eliminated the change in SNGFR achieved changing salt content of the perfusate (ΔSNGFR = –0.1 ± 1.6 vs. 11.2 ± 2.6 nL/min, clamp versus control, P = 0.001). In other words, clamping ADO activity abolished the TGF–mediated change in SNGFR during manipulation of macula densa salt by retrograde perfusion.

Figure 7

SNGFR by paired proximal collections during retrograde perfusion of Henle’s loop from the early distal tubule. Perfusion was at 12.5 nL/min with artificial tubular fluid (ATF) containing either 10 or 50 mM NaCl to effect minimum or maximum stimulation of TGF. To clamp adenosine activity, 5′-nucleotidase blocker and cyclohexyladenosine were added to tubular ATF. P values refer to effect of changing salt concentration. ^ABy two-way ANOVA, the adenosine clamp reduced the TGF response (P = 0.0006) but had no effect on SNGFR independent of TGF (P = 0.75).

The present studies imply that adenosine exerts a tonic influence on the glomerular microvasculature and contributes to the compensatory changes in SNGFR that occur when flow is altered in the tubule. Evidence that 5′-nucleotidase activity influences basal glomerular hemodynamics is provided by two sets of experiments. First, delivering the 5′-nucleotidase inhibitor, MADP, into the free-flowing late proximal tubule caused baseline tubular flow to increase, suggesting that SNGFR had increased. A similar response was obtained with the adenosine A1 receptor antagonist, KW3902, and an opposite response was obtained with the adenosine A1 agonist, CHA. The role of 5′-nucleotidase activity as a determinant of SNGFR was confirmed in a second set of studies in which SNGFR was measured directly and found to be increased during perfusion of the loop of Henle with MADP, irrespective of the perfusion rate (Figures 3 and 4).

In addition to influencing basal SNGFR, adenosine formed by 5′-nucleotidase was confirmed to participate in TGF. Inhibiting 5′-nucleotidase with MADP reduced the homeostatic efficiency of TGF as manifest by a lesser fractional compensation for small perturbations in V_LP. Fractional compensation is the integrated result of glomerulotubular balance and TGF. After adjusting for the contribution of GTB to overall TGF efficiency (Figure 5), MADP reduced the maximum slope of the TGF curve by more than 80% (Figure 6). MADP also reduced the maximum range of the TGF response by 35–45% (Figure 4).

Autacoid or paracrine substances that impinge on TGF may either mediate or modulate TGF. To qualify as a TGF mediator, a substance must meet the following two criteria. First, its activity must change in response to changes in delivery of NaCl to the luminal macula densa, and these changes must occur within the minute-to-minute time frame of the TGF response. Second, assuming there is only one TGF mediator, preventing the activity of that mediator from changing in response to signals from the macula densa will abolish the TGF response. In contrast, a TGF modulator does not actually cause the TGF response but operates in the background to condition the response. The activity of a modulator might or might not be linked to changes in macula densa NaCl, but if such a link is present, interrupting the link will not abolish the TGF response. It is common to think of a modulator as something that reacts more slowly than the system it modulates. However, a system that responds rapidly to changes in macula densa NaCl is not excluded from being a TGF modulator as long as there is some level of activity at which the putative modulator can be clamped without suppressing the TGF response.

In practice, it is simpler to demonstrate that a substance modulates TGF than it is to demonstrate that a substance mediates TGF, and several candidate mediators have been reduced to modulators. For example, several characteristics of the renin-angiotensin system make angiotensin II a candidate mediator for TGF. However, it has been demonstrated by these criteria that angiotensin II is an important TGF modulator but does not mediate TGF (18). Similarly, the abundance of nitric oxide synthase in the macula densa combined with the vasodilatory influence of nitric oxide on the glomerular microvasculature suggests that TGF might be mediated by an inverse dependence of nitric oxide generation on macula densa NaCl. However, it has also been demonstrated that macula densa nitric oxide is a modulator, not a mediator, of TGF (reviewed in ref. 19).

In the present case, inhibiting 5′-nucleotidase reduced both the slope of the TGF curve and the maximum range of the TGF response. This is sufficient to prove that 5′-nucleotidase impinges on the TGF system, but it is not sufficient to determine whether adenosine is a mediator or a modulator of TGF. In fact, the effects of MADP could be accounted for if adenosine were merely operating in the background to intensify the response to some other vasoconstrictor or vasodilator. To test the hypothesis that adenosine is an actual mediator of TGF, experiments were performed to determine whether restoring adenosine activity to pre-MADP levels with exogenous adenosine A1 agonist would reverse the effects of MADP on the TGF response. This was the idea behind the ADO-clamp experiments in which the adenosine A1 receptor agonist, CHA, was administered along with the 5′-nucleotidase inhibitor, MADP. If the effects of MADP on TGF were reversed by addition of CHA, then adenosine would be an unlikely mediator of TGF. On the other hand, if the effects of MADP on the TGF response were not reversed by adding CHA, this would be evidence that the mere presence of adenosine is not sufficient to account for its role in TGF. Instead, adenosine activity must be allowed to fluctuate along with changes in tubular flow for TGF to operate normally, as would be expected if adenosine were a mediator of TGF.

In practice it is not possible to clamp adenosine activity perfectly at pre-MADP levels because that level will be different in every nephron. Furthermore, it is not possible to know the degree to which MADP prevents adenosine from fluctuating when the TGF signal is manipulated unless MADP were to eliminate TGF completely. However, to prove that the actions of TGF can be suppressed by clamping adenosine activity, it is not essential that the clamp be perfect or that adenosine activity be identical to baseline. To achieve positive proof that adenosine mediates TGF, the model must meet two less-stringent conditions. The first condition is that the clamp must be tight enough to alter the TGF response measurably. The second condition is that the clamped adenosine activity should not be less than baseline but must not be so high as to potentially overwhelm the vascular response to other unknown TGF mediators. In the present experiments, the first condition was clearly met, as the ADO clamp had effects on the behavior of TGF that were similar to the effect of MADP alone (Figures 2 and 4).

The second condition also appears to have been met, although this is less obvious. As surrogate measurements of adenosine activity in the different experiments, measurements were made of unperturbed V_LP or SNGFRmean. In the free-flow experiments, enough CHA was given to reduce V_LP by approximately 4 nL/min during the ADO clamp (Figure 1). Therefore, it cannot be argued that too little CHA was given in these experiments. However, it is possible that too much CHA was given and that the associated decrease in ambient tubular flow caused the ambient flow to shift from a steep to a flat portion of the TGF curve. Alone, this could account for the apparent loss of fractional compensation, as fractional compensation depends on the slope of the TGF curve over the range in which the nephron operates during the perturbation experiment (see Eq. 6). However, when CHA was added alone, V_LP decreased by 6 nL/min, yet fractional compensation was unaffected (Figures 1 and 2). Therefore, it is unlikely that the ADO clamp lessened the fractional compensation purely by altering the position of the operating point along the TGF curve.

In the wax-block experiments, relatively less CHA was used. In these experiments, SNGFRmean was used as the surrogate assay for the underlying adenosine activity. MADP alone caused SNGFRmean to increase significantly, and this effect was completely reversed by the ADO clamp (Figure 4). Therefore, adenosine activity achieved in these experiments met the conditions required to prove that TGF will not function normally unless adenosine activity is allowed to fluctuate from minute to minute. The impact of clamping ADO activity on the TGF response was even more apparent during retrograde perfusion, which provided a shorter path for delivering drugs to the macula densa. In this case, clamping ADO activity appeared to abolish TGF altogether, although the applied stimulus (10–50 mM NaCl variation) should dissipate somewhat before reaching the macula densa and is less than the range of macula densa NaCl variation expected during 4–38 nL/min perfusion from the late proximal tubule (11).

It was not the main purpose of this study to delineate the role of 5′-nucleotidase or adenosine in normal function of the proximal tubule. However, to extract the slope of the TGF curve from the response to flow perturbations in free-flowing nephrons, it was necessary to know how glomerular tubular balance was affected when MADP or MADP + CHA was perfused through the loop of Henle. Both MADP and MADP + CHA somewhat reduced the dependence of proximal reabsorption on SNGFR, and the addition of CHA to MADP reduced proximal reabsorption relative to MADP alone (Figure 5). From the current experiments, it cannot be determined whether the effects of MADP and CHA on the proximal tubule were direct or indirect. Because of the experimental design, the drugs could only arrive at the early proximal tubule circuitously and proximal reabsorption might have been affected by differences in peritubular oncotic pressure arising from impacts of CHA and/or MADP on glomerular filtration fraction. Nonetheless, the difference in proximal reabsorption between nephrons perfused with MADP alone and those perfused with MADP + CHA suggests that adenosine, acting through A1 receptors can reduce proximal reabsorption. Furthermore, the finding of impaired glomerular tubular balance in nephrons perfused with either MADP or MADP + CHA suggests that the modulation of local adenosine concentration by 5′-nucleotidase might be necessary for normal glomerular tubular balance. However, there is no evidence from the present findings that proximal tubular reabsorption is autoregulated by transport-dependent adenosine formation (Figure 5, center panel). Whereas proximal reabsorption was affected by adenosine analogs delivered into the loop of Henle, there was no effect of these drugs on loop of Henle reabsorption, per se.

The current data imply that adenosine that is generated by 5′-nucleotidase mediates a compensatory change in SNGFR when flow is altered beyond the late proximal tubule. However, these data do not prove that the substrate for this 5′-nucleotidase comes from ATP hydrolyzed by macula densa cells. In fact, it is known that macula densa cells exhibit relatively little Na-K-ATPase activity relative to other cells in the neighboring thick ascending limb (20). This raises the possibility that tubular-glomerular interactions mediated by adenosine do not involve the macula densa in the same way as usually envisioned for TGF. On the other hand, relative changes in the abundance AMP that result from changing rates of ATP hydrolysis depend strongly on the ATP/AMP ratio and not merely on the absolute rate of ATP hydrolysis. This can be demonstrated mathematically as follows.

In cells, ATP, ADP, and AMP are maintained in equilibrium by the following reaction (Rxn 1), which is mediated by adenylate kinase (21): ATP + AMP ↔ 2ADP. Accordingly,

(Equation 10)10

where k is the equilibrium constant. If the system is essentially closed with respect to total nucleotide content while the number of high-energy phosphate bonds is allowed to change through hydrolysis and formation of ATP, then ATP, ADP, and AMP can change, but

(Equation 11)11

When ATP is hydrolyzed to ADP, some of this ADP will be converted to ATP + AMP according to Rxn 1. Therefore, AMP formation will track ATP hydrolysis and a gain for this process can be defined as:

(Equation 12)12

Expanding Eq. 11 and substituting from Eq. 10 and Eq. 11 gives

(Equation 13)13

Rearranging Eq. 13 and dividing through by AMP/ATP gives

(Equation 14)14

This is a monotonically increasing function of ATP/AMP. Therefore, fractional changes in AMP are a more sensitive signal of changes in ATP in cells where the starting ATP/AMP ratio is high. Therefore, the transduction of changes in luminal NaCl delivery into changes in AMP formation might have high gain in the macula densa, even though the absolute transport rate by these cells is low.

Adenosine is generated in the JGA by 5′-nucleotidase and exerts a tonic influence on single nephron glomerular filtration rate and on the process of TGF. Both the maximum range over which TGF can cause SNGFR to change and the strength of the TGF response to small perturbations in ambient tubular flow are reduced when adenosine formation is suppressed or when adenosine activity is clamped. These findings establish adenosine as a mediator of TGF.

This work was performed with funds provided by the Department of Veterans Affairs Research Service and by the NIH (National Institute of Diabetes and Digestive and Kidney Diseases, RO1 DK-28602). V. Vallon was supported by grants from the Deutsche Forschungsgemeinschaft (Va 118/3-2) and the Federal Ministry of Education and Research (BMBF 01 EC 9405 and 01 KS 9602). KW-3902 was kindly provided by H. Osswald (Department of Pharmacology, University of Tuebingen, Tuebingen, Germany).

1. Osswald, H, Muhlbauer, B, Schenk, F. Adenosine mediates tubuloglomerular feedback response: an element of metabolic control of kidney function. Kidney Int 1991. 39:S128-S131.
2. Schnermann, J, Weihprecht, H, Briggs, J. Inhibition of tubuloglomerular feedback during adenosine-1 receptor blockade. Am J Physiol 1990. 258:F553-F561.
   View this article via: PubMed Google Scholar
3. Weihprecht, H, Lorenz, JN, Briggs, JP, Schnermann, J. Synergistic effects of angiotensin and adenosine in the renal microvasculature. Am J Physiol 1994. 266:F227-F239.
   View this article via: PubMed Google Scholar
4. Pflueger, AC, Schenk, F, Osswald, H. Increased sensitivity of the renal vasculature to adenosine in streptozotocin-induced diabetes mellitus rats. Am J Physiol 1995. 269:F529-F535.
   View this article via: PubMed Google Scholar
5. Thomson, S, Vallon, V, Blantz, RC. Asymmetry of tubuloglomerular feedback effector mechanism with respect to ambient tubular flow. Am J Physiol 1996. 271:F1123-F1130.
   View this article via: PubMed Google Scholar
6. Huang, WC, Bell, PD, Harvey, D, Mitchell, KD, Navar, LG. Angiotensin influences on tubuloglomerular feedback mechanism in hypertensive rats. Kidney Int 1988. 34:631-637.
   View this article via: PubMed CrossRef Google Scholar
7. Schnermann, J, Briggs, JP. Restoration of tubuloglomerular feedback in volume-expanded rats by angiotensin II. Am J Physiol 1990. 259:F565-F572.
   View this article via: PubMed Google Scholar
8. Mitchell, KD, Navar, LG. Enhanced tubuloglomerular feedback during peritubular infusions of angiotensins I and II. Am J Physiol 1988. 255:F383-F390.
   View this article via: PubMed Google Scholar
9. Blantz, R.C., and Tucker, B.J. 1978. Measurements of glomerular dynamics. In Methods in pharmacology. M. Martinez-Maldonado, editor. Plenum Publishing Co. New York, New York, USA. 141–163.
   View this article via: PubMed Google Scholar
10. Thomson, SC, Blantz, RC. Homeostatic efficiency of tubuloglomerular feedback in hydropenia, euvolemia, and acute volume expansion. Am J Physiol 1993. 264:F930-F936.
    View this article via: PubMed Google Scholar
11. Vallon, V, Thomson, SC. Inhibition of local nitric oxide synthase increases the homeostatic efficiency of tubuloglomerular feedback. Am J Physiol 1995. 38:F892-F899.
12. Vallon, V, Blantz, RC, Thomson, S. Homeostatic efficiency of tubuloglomerular feedback is reduced in established diabetes mellitus in rats. Am J Physiol 1995. 38:F876-F883.
13. Thomson, SC, Blantz, RC, Vallon, V. Increased tubular flow induces resetting of tubuloglomerular feedback in euvolemic rats. Am J Physiol 1996. 270:F461-F468.
    View this article via: PubMed Google Scholar
14. Thomson, SC, Vallon, V, Blantz, RC. Reduced proximal reabsorption resets tubuloglomerular feedback in euvolemic rats. Am J Physiol 1997. 273:R1414-R1420.
    View this article via: PubMed Google Scholar
15. Schnermann, J., and Briggs, J. 1985. Function of the juxtaglomerular apparatus: local control of glomerular hemodynamics. In The kidney: physiology and pathophysiology. D.W. Seldin and G. Giebisch, editors. Raven Press. New York, New York, USA. 669–697.
    View this article via: PubMed Google Scholar
16. Vallon, V, Osswald, H, Blantz, RC, Thomson, S. Potential role of luminal potassium in tubuloglomerular feedback. J Am Soc Nephrol 1997. 8:1831-1837.
    View this article via: PubMed Google Scholar
17. Vallon, V, Richter, K, Blantz, RC, Thomson, S, Osswald, H. Glomerular hyperfiltration in experimental diabetes mellitus: potential role of tubular reabsorption. J Am Soc Nephrol 1999. 10:2569-2576.
    View this article via: PubMed Google Scholar
18. Schnermann, J, Briggs, J. Restoration of tubuloglomerular feedback in volume-expanded rats by angiotensin II. Am J Physiol 1984. 247:F246-F251.
    View this article via: PubMed Google Scholar
19. Welch, WJ, Wilcox, CS, Thomson, SC. Nitric oxide and tubuloglomerular feedback. Semin Nephrol 1999. 19:251-262.
    View this article via: PubMed Google Scholar
20. Schnermann, J, Marver, D. ATPase activity in macula densa cells of the rabbit kidney. Pflugers Arch 1986. 407:82-86.
    View this article via: PubMed CrossRef Google Scholar
21. Mayes, P. 1979. Bioenergetics. In Review of physiological chemistry. H.A. Harper, V.W. Rodwell, and P.A. Mayes, editors. Lange Medical Publications. Los Altos, California, USA. 260–265.
    View this article via: PubMed Google Scholar