Prostaglandin E2 binding sites in bovine iris-ciliary body.

In the eye, prostaglandins (PGs), in particular PGE2 and PGF2 alpha, may induce vasodilation, disruption of the blood-aqueous barrier, and biphasic effects on intraocular pressure, depending on the species. The initial event leading to many of these physiologic responses is the interaction between the PG and a receptor. We have explored the specificity and selectivity of PGE2 receptors in bovine iris-ciliary body (ICB) membrane preparations. Pigment-free bovine ICB membranes were prepared by high-speed sucrose density-gradient centrifugation. Membranes were incubated with 1 nM 3H-PGE2 in the presence or absence of varying concentrations of unlabeled PGE2 or F2 alpha. Binding of 3H-PGE2 to membranes at 37 degrees C increased linearly with protein concentration, and binding reached equilibrium in 30 min. Specific PGE2 binding represented 80% of total 3H-PGE2 binding. Studies with unlabeled PGE2 or F2 alpha, as competing ligands, showed a dose-dependent inhibition of 3H-PGE2 specific binding. The IC50 for unlabeled PGE2 and F2 alpha was 3 and 379 nM, respectively, which suggests a 100-fold greater selectivity of the binding sites for PGE2 over F2 alpha. Scatchard analysis of saturation data revealed a mean Kd value of 13.3 nM with a Bmax of 156 fmoles bound/mg protein. The general linearity of our Scatchard plots tends to suggests a single class of binding sites for PGE2, although more than a single binding site could be present. These results indicate that binding sites selective for PGE2 exist in the bovine ICB.

Immunocytochemical demonstration of vasopressin binding in rat kidney.

We investigated the immunoperoxidase demonstration of vasopressin (VSP) bound to paraffin-embedded sections of rat kidney and the effects of various fixatives. Slices of rat kidney from normal and 4-day water-deprived rats were incubated with 10(-7) M VSP, fixed, and embedded in paraffin. Hydrated sections of these tissues were again incubated with 10(-7) M VSP or 10(-7) M VSP and 10(-5) M oxytocin (OXY). VSP bound to the sections was demonstrated using rabbit anti-Arg8 VSP antiserum and peroxidase-labeled second antibody. In sections of kidney from both normal and water-deprived rats, immunoperoxidase labeling was most intense in the renal papilla and was restricted to the cells of the ducts of Bellini and loops of Henle. In the medulla, the collecting ducts and medullary thick ascending limbs of Henle were moderately stained. In the normal kidney sections there was no staining of the proximal tubules, distal convoluted tubules (DCT), and only slight staining of the cortical collecting ducts (CCD). However, in the water-deprived rats there was a considerable increase in the staining of the DCT and CCD. Simultaneous incubation in OXY and VSP resulted in reduced immunoperoxidase labeling of the tubules. Omission of VSP incubation led to a similar decrease in stain intensity, indicating a specificity for the sites of VSP binding. This technique allows the identification of cells responsible for the binding of VSP in the kidney.

Vasopressin receptors in human seminal vesicles: identification, pharmacologic characterization, and comparison with the vasopressin receptors present in the human kidney.

Because of the presence of a high density of vasopressin receptors in the epithelial cells of porcine seminal vesicles similar to the V2 vasopressin receptors of renal tubules, human seminal vesicles and kidney were investigated using quantitative binding and adenylate cyclase studies. Tissues were obtained at surgery from 17 patients with urologic diseases. A homogeneous class of vasopressin binding sites have been found in both seminal vesicles and renal medulla. However, the vasopressin receptors present in these tissues are different in terms of ligand specificity and adenylate cyclase activation. In seminal vesicles, the V1 vasopressin antagonist d(CH2)5 TyrMeAVP is 36-fold, more potent than the V2 agonist dVDAVP in displacing [3H]AVP binding, while in the medullopapillary portion of kidney dVDAVP is 24-fold, more selective than d(CH2)5 TyrMeAVP for the arginine vasopressin binding site. Furthermore, arginine vasopressin induces a dose-dependent increase in adenylate cyclase activity in renal membranes, while it was ineffective in seminal vesicle membranes. These results indicate that a very high affinity (0.2 nM), low capacity (14 fmoles/mg protein) class of vasopressin receptors is present in human seminal vesicles, having pharmacologic characteristics similar to the V1 subtype of vasopressin receptors. The presence of a high affinity (1.6 nM), high capacity (350 fmoles/mg protein) V2 subtype of vasopressin receptors in human renal membranes is also confirmed. The density of the vasopressin receptors present in human seminal vesicles is inversely correlated with patient age, consistent with a physiologic role for vasopressin in the regulation of accessory sex gland activity.

Sorbitol metabolism in inner medullary collecting duct cells of diabetic rats.

Intracellular accumulation of sorbitol, generated from D-glucose via the aldose reductase pathway, is thought to play an important role in diabetic complications such as lens cataracts and neuropathy. In order to elucidate the effect of diabetes on the renal inner medulla, another sorbitol-rich tissue, male Wistar rats were treated with a single dose of streptozotocin (60 mg/kg body weight, i.p.). Six weeks later total inner medullary tissue (IM) or isolated inner medullary collecting duct (IMCD) cells were prepared. In diabetic IM tissue, sorbitol content was 1.8-fold higher than in control IM tissue (134 +/- 17 vs. 74 +/- 22 mumol/g tissue protein). Sorbitol production in both normal and diabetic IMCD cells was strongly dependent on extracellular D-glucose concentration. In normal cells, for example, sorbitol production was 90 +/- 9 mumol sorbitol/g protein x h at 45 mM D-glucose compared to 13 +/- 1 mumol/g protein x h at 5 mM. At identical D-glucose concentrations sorbitol synthesis in diabetic IMCD cells was, however, always significantly higher than in control cells (122% of control at 15 mM and 126% of control at 45 mM). In addition, aldose reductase activity in diabetic IM was found to be augmented. The maximal velocity was 4.2 times higher (97 +/- 22 U/g protein vs. 23 +/- 7 U/g protein) while the Km of the enzyme remained unchanged. Membrane permeability for sorbitol or the response to changes in extracellular osmolarity was not significantly different in diabetic IMCD cells and normal cells with correspondingly high intracellular sorbitol concentrations. Similarly the kinetic parameters of D-glucose uptake were not altered by streptozotocin treatment.(ABSTRACT TRUNCATED AT 250 WORDS)

Renal proton magnetic resonance in experimental acute renal failure in rats.

Cortical, medullary and papillary T1 and T2 water proton relaxation times were measured at 37 degrees C, 20 MHz. The measurements were made using kidneys from rats affected by many forms of experimental acute renal failure (ARF), namely acute hemorrhagic hypovolemia, angiotensin II administration, antidiuretic hormone (ADH) administration, glycerol, and other nephrotoxins (gentamicin, cisplatinum, cyclosporine), renal artery occlusion for different periods of time, and ureteral ligation. From the T1 and PW (percent tissue water content) the bound water (FB) and HF (percent water bound/g solid) were calculated according to a fast proton diffusion model. In most experimental models studied, the experiments were repeated following paramagnetic enhancement with GdDTPA administration (70 mmol/kg BW). By profiling the deviations from normal, it was possible to differentiate the ischemic (shortened T1, prolonged T2), obstructive (very high T1 and T2 in both cortex and medulla) and nephrotoxic (prolonged T2) forms of ARF. Significant changes in free/bound water compartments occurred, though their biological significance is unknown. T1 and T2 ratios before and after paramagnetic enhancement correlated well with estimates of glomerular filtration rate. In the first minutes following acute hemorrhagic hypovolemia, the intrarenal water distribution remained unchanged. After GdDTPA significant water proton T1 and T2 changes characterized the immediate posthemorrhagic state similar to the effect of ADH.

Regional distribution of immunoreactive endothelin in porcine tissue: abundance in inner medulla of kidney.

A specific and sensitive radioimmunoassay for endothelin has been developed. Half maximal inhibition of binding of radioiodinated endothelin was observed at 37 pg/tube and endothelin was detectable as low as 1 pg/tube. With this assay, the regional distribution of endothelin was determined in porcine tissue. The highest concentration of immunoreactive endothelin was observed in inner medulla of kidney (6.2 +/- 1.1 pg/mg wet weight), while the concentration in kidney cortex was very low. Immunoreactive endothelin was also found in lung in relatively high concentration. The immunoreactive endothelin in porcine lung and inner medulla of kidney was further characterized by reverse phase high performance liquid chromatography combined with radioimmunoassay.

A simple HPLC method for quantitating major organic solutes of renal medulla.

A simple high performance liquid chromatography (HPLC) method of separating and quantitating the predominant organic solutes of the renal medulla is described. These organic solutes include myo-inositol, glycerophosphorylcholine, sorbitol, betaine, and urea. Other physiologically significant solutes, including glucose and mannitol, can be separated and quantitated concurrently with this method. With the use of this technique, the organic solutes of the rabbit kidney were determined. No new organic compounds were detected by HPLC that could significantly contribute to intracellular osmolality of the medulla. The values for the organic solutes already described were similar to those obtained by more complicated and limited approaches such as classical enzymatic techniques, ion electrodes, nuclear magnetic resonance spectroscopy, and gas chromatography-mass spectroscopy.

Postmortem metabolic indices of antemortem adrenergic stimulation.

Tissue lactate concentration has been reported to be a useful postmortem indicator of antemortem awareness of mortal danger. The purpose of this study was to determine further whether selected tissue metabolites could be used as postmortem markers of antemortem adrenergic stress. Sprague-Dawley albino rats were anesthetized with pentobarbital and then injected with 2.0 mg kg-1 i.p. epinephrine hydrochloride to induce experimentally a severe sympathetic response that may be associated with the awareness of mortal danger; 20 min after the injection of epinephrine, when the metabolic response was at its peak, the animals were killed by exsanguination. Samples of the following tissues were removed immediately prior to death (0 h) and 48 h postmortem: soleus, plantaris, kidney medulla, kidney cortex, liver, and heart. These samples were analyzed for glycogen, lactate, ATP, creatine phosphate, pH, and total protein concentration. Significant differences in lactate concentration were observed in all tissues except soleus at 0 h in the epinephrine-injected animals. Specific tissues also had significant reductions in glycogen, ATP, and creatine phosphate concentrations at 0 h. At 48 h postmortem, however, only the liver and soleus lactate concentrations were significantly different from the 48-h control samples. It is unlikely that these small differences found in some tissues at 48 h postmortem would be detected in an uncontrolled accident situation. We concluded from these findings that these selected tissue metabolites are not useful as long-term postmortem indicators of antemortem adrenergically induced hypermetabolism.

Solubilization of a guanine nucleotide-sensitive form of vasopressin V2 receptors from porcine kidney.

Vasopressin (V2) receptors were solubilized from porcine kidney membranes with the detergent egg lysolecithin. Binding of [3H]vasopressin to the solubilized fraction was rapid, specific, and saturable. The agonist dissociation constants observed in membranes and solubilized fractions were 1.7 +/- 0.3 and 2.3 +/- 0.2 nM, respectively. In competition binding experiments, the solubilized fraction exhibited the same pharmacological profile as the membranes. Chemical crosslinking of [125I]vasopressin to the solubilized fraction followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis demonstrated a 62-kDa band which was specifically labeled with [125I]vasopressin. Vasopressin binding sites from the solubilized fractions were resolved by gel filtration and ultracentrifugation on a sucrose gradient. In addition, agonist high affinity binding to V2 receptors and its sensitivity to guanine nucleotides were preserved even after solubilization in the absence of prebound agonist prior to solubilization. Addition of guanine nucleotides such as GTP gamma S decreased the specific binding of [3H]arginine vasopressin to these solubilized fractions in a dose-dependent manner, suggesting the solubilization of a V2 receptor-G protein complex. [32P]ADP ribosylation of the solubilized fraction by cholera and pertussis toxins revealed specifically labeled proteins with molecular weights of 42,000-43,000 and 39,000-41,000, respectively, on sodium dodecyl sulfate polyacrylamide gels. Furthermore [35S]GTP gamma S binding to these solubilized fractions was enhanced by vasopressin, confirming that a significant proportion of the vasopressin receptors must be closely coupled to G proteins even when these receptors are solubilized in the absence of agonist. These results are in contrast with those reported for beta, alpha 2 adrenergic and D2 dopaminergic receptor systems, but in agreement with D1 dopaminergic and A1 adenosine receptors. The molecular mechanism responsible for this difference remains to be determined.

Cholinergic receptors in renal medullary collecting duct cells.

Intrarenal administration of cholinergic agents produces diuresis. However, neither cholinergic innervation or specific cholinergic receptors have been shown to be present in the kidney. Recently, we have demonstrated that carbachol, a cholinergic agent, stimulates phosphoinositide hydrolysis in the inner medullary collecting duct (IMCD) cells. The effect was blocked by atropine (a cholinergic antagonist), suggesting that phosphoinositide hydrolysis occurs through the interaction of carbachol with specific cholinergic receptors in these cells. Therefore, we examined the cholinergic receptors in IMCD cells by measurement of radioligand binding of a cholinergic receptor antagonist, I-quinuclidinyl (phenyl-4-3H)benzilate([3H]QNB). The IMCD cells were prepared from rabbit kidneys by incubating the inner medullary slices with collagenase and treating the isolated cells with hypotonic solution to lyse cells other than IMCD cells. Binding of [3H]QNB to IMCD cells was measured at 37 degrees C for 60 min in the absence (total binding) and the presence (nonspecific binding) of 100 microM atropine (a muscarinic receptor antagonist). The specific binding (the difference between total and nonspecific binding) of [3H]QNB to IMCD cells was saturable with a Bmax (maximum binding sites) of 27.5 fmol/mg of protein and Kd (dissociation constant) of 0.27 nM. Atropine, but not hexamethonium (a nicotinic antagonist), was able to displace [3H]QNB from IMCD cells with a Ki of 0.1 microM. It is, therefore, concluded that specific high affinity muscarinic receptors are present in IMCD cells. These receptors may play a role in producing the pharmacologic actions of cholinergic agents on the kidney.

Determination of doxycycline in bovine tissues and body fluids by high-performance liquid chromatography using photodiode array ultraviolet-visible detection.

A sensitive and selective high-performance liquid chromatographic assay was developed to quantitate doxycycline concentrations in bovine tissues and body fluids. The method involved sonication of doxycycline-spiked minced tissues with appropriate solutions to extract tissue-bound drug. Acetonitrile:85% phosphoric acid:water (20:2:78) was added to doxycycline-spiked serum or urine. The mixtures were ultrafiltered through 30,000 or 10,000 Da molecular weight cutoff microseparation systems. Separation was obtained by a reversed-phase microbore column. Absorbance of the column effluent was measured by an ultraviolet-visible photodiode array detector scanning from 235 to 380 nm and/or a photometric detector operated at 268 or 345 nm. Chromatographic peak homogeneity was evaluated by three-dimensional spectrochromatograms, contour maps, and absorbance ratios.

Verotoxin receptor glycolipid in human renal tissue.

Infection with verotoxin producing Escherichia coli has been strongly implicated in the etiology of the hemolytic uremic syndrome (HUS). We have previously shown that this toxin specifically binds to a glycolipid receptor-globotriosyl ceramide (Gb3). We have therefore quantitated the level of this glycolipid by HPLC in human renal cortex and medulla as a function of age. We have also measured the binding of verotoxin to Gb3 isolated from each renal tissue sample. Gb3 was a major component of the glycolipid fraction of all renal samples analyzed. The levels were found to be higher in the cortex than medulla, correlating with the clinical incidence of renal lesions in HUS, but reduced in the kidneys of infants as compared to adults. Verotoxin binding was directly proportional to the renal Gb3 content. Thus, human renal tissue is a rich source of the verotoxin receptor glycolipid. However, changes in receptor concentration cannot explain the age-related incidence of HUS.

Element composition of tubule cells in the inner stripe of the renal outer medulla.

To obtain further insight into renal medullary function, element concentrations were determined in individual tubule cells of the outer medulla in the rat kidney using electron microprobe analysis on freeze-dried cryosections. In the cells of the thick ascending limb of Henle's loop the Na, P, Cl, and K concentrations (means +/- SEM) were: 9.5 +/- 0.6, 158.4 +/- 6.2, 25.6 +/- 1.2, and 135.3 +/- 4.8 mmol/kg wet weight, respectively. While similar Na, P, and K concentrations were observed in light and dark cells of the medullary collecting duct, Cl was markedly higher--55.0 +/- 2.8 mmol/kg wet weight--in the dark cells. The electrolyte concentrations of the thick ascending limb cells seen in the present study are in good agreement with ion activities reported for the isolated perfused thick ascending limb. The low cell Na and Cl concentrations provide a favorable driving force for passive cell entry of Na, Cl, and K across the apical membrane via the Na-2Cl-K cotransporter even at low tubule fluid NaCl concentrations. Although in hydropenic rats interstitial tonicity of the inner stripe is above isotonicity, electrolyte concentrations of inner stripe cells did not differ from those obtained in cortical tubule cells. This finding suggests that, similar to papillary cells, osmoadaptation of outer medullary cells is, at least partially, accomplished by organic osmolytes.

Glycosaminoglycans of the rat renomedullary interstitium: ultrastructural and biochemical observations.

The rat renal papillary interstitum which contains abundant proteoglycans is a unique area important in renal function. These proteoglycans were studied ultrastructurally by ruthenium red fixation and staining and phosphate-buffered fixation before and after enzyme digestion. A tissue culture of rat renomedullary interstitial cells, the predominant cell of the renal papillary interstitum, was studied for its ability to synthesize proteoglycans and the proteoglycans were then analyzed. Tissue slices of whole rat renal inner medulla were also evaluated for their synthetic ability. In combination, these studies indicate that the dominant glycosaminoglycan is hyaluronic acid. The tissue culture of rat renal medullary interstitial cells synthesized glycosaminoglycans and on analysis, hyaluronic acid was found to be the chief glycosaminoglycan secreted by the renomedullary interstitial cells. Combined with the removal of the proteoglycans from tissue by leech hyaluronidase and testicular hyaluronidase, this suggests that the dominant glycosaminoglycan is hyaluronic acid. Hyaluronic acid is also synthesized by the intact papilla confirming the findings with the tissue culture. However, in addition, sulfated glycosaminoglycans were also synthesized by the intact papilla, presumably the product of the noninterstitial components of the papilla.

High- and low-affinity binding sites for vasoactive intestinal peptide (VIP) in the rat kidney revealed by light microscopic autoradiography.

The distribution of VIP binding sites in rat kidney and adrenal gland has been examined by light microscopic autoradiography. A fully characterized mono-iodinated molecular form of VIP (M-125-I-VIP) which maintains the biological activity of the native peptide, was used for this study. Two types of VIP binding sites, with high and low affinity, have been identified. High affinity sites are associated with (i) glomerular structures in the cortex, (ii) the inner stripe of the outer medulla, possibly corresponding to Henle's loops and distal tubules, (iii) radiated structures in the inner zone of the medulla, likely to represent labeling of collecting ducts and/or vascular bundles and (iv) the adrenal cortex. Autoradiographic grains associated with low affinity sites are present diffusely throughout the renal cortex, possibly corresponding to labeling of tubular and/or vascular structures, and throughout the adrenal gland. These observations further delineate a role of VIP in renal and neuroendocrine function.

Relation between the anion exchange protein in kidney medullary collecting duct cells and red cell band 3.

A membrane protein that is immunochemically similar to the red cell anion exchange protein, band 3, has been identified on the basolateral face of the outer medullary collecting duct (MCD) cells in rabbit kidney. In freshly prepared separated rabbit MCD cells, M.L. Zeidel, P. Silva and J.L. Seifter (J. Clin. Invest. 77:1682-1688, 1986) found that C1-/HCO-3 exchange was inhibited by the stilbene anion exchange inhibitor, DIDS (4,4'-diisothiocyano-2,2'-disulfonic stilbene), with a K1 similar to that for the red cell. We have measured the binding affinities of a fluorescent stilbene inhibitor, DBDS (4,4'-dibenzamido-2,2'-disulfonic stilbene), to MCD cells in 28.5 mM citrate and have characterized both a high-affinity site (Ks1 = 93 +/- 24 nM) and a lower affinity site (Ks2 = 430 +/- 260 nM), which are closely similar to values for the red cell of 110 +/- 51 nM for the high-affinity site and 980 +/- 200 nM for the lower affinity site (A.S. Verkman, J.A. Dix & A.K. Solomon, J. Gen. Physiol. 81:421-449, 1983). When Cl- replaces citrate in the buffer, the two sites collapse into a single one with Ks1 = 1500 +/- 400 nM, similar to the single Ks1 = 1200 +/- 200 nM in the red cell (J.A. Dix, A.S. Verkman & A.K. Solomon, J. Membrane Biol. 89:211-223, 1986). The kinetics of DBDS binding to MCD cells at 0.25 microM-1 are characterized by a fast process, tau = 0.14 +/- 0.03 sec, similar to tau = 0.12 +/- 0.03 sec in the red cell. These similarities show that the physical chemical characteristics of stilbene inhibitor binding to MCD cell 'band 3' closely resemble those for red cell band 3, which suggests that the molecular structure is highly conserved.

C-reactive protein-mediated complement activation in polymyalgia rheumatica and other systemic inflammatory diseases.

An immunoglobulin-independent deposition of the complement (C) components C4 and C3 occurs on rat kidney medullary structures, when sera of patients with various inflammatory diseases are studied by indirect immunofluorescence. The diagnostic value of this new test (C4-C3-IFT) for polymyalgia rheumatica (PMR) is stressed, since all sera from patients with active disease yielded positive reactions. Though highly sensitive with respect to PMR, C4/C3-IFT is not specific for this syndrome. Examples of positive reactions in systemic inflammatory diseases other than PMR are documented. Besides the clinical studies, C4/C3-IFT reactivity was analyzed with regard to the mechanisms of the reaction. Experimental data are presented which suggest that C-reactive protein (CRP) binds to rat kidney structures, thereby activating the classical C cascade. As a result of CRP-C interaction, C4 and C3 components are fixed to distinct renal medullary structures. Because of its technical simplicity, C4/C3-IFT can routinely be used to screen patients' sera for CRP-mediated C activation. This ex vivo test system may contribute to a better understanding of pathophysiological functions of serum CRP in various inflammatory diseases.

Effect of protein intake and urea on sodium excretion during inappropriate antidiuresis in rats.

Administration of urea to patients with the syndrome of inappropriate antidiuresis (SIAD) is thought to ameliorate hyponatremia by both producing an osmotic diuresis and diminishing ongoing natriuresis. The present study evaluated these effects in a rat model of SIAD utilizing dilutional hyponatremia induced by continuous infusion of 1-deamino-[8-D-arginine] vasopressin. Following 48 hours of sustained hyponatremia, separate groups of rats were then refed with either: (1) 5% dextrose alone, (2) a 20% protein chow, (3) an isocaloric protein deficient (0%) chow, or (4) the isocaloric protein-deficient chow supplemented with oral urea. Our results demonstrate that rats refed a 20% protein diet significantly improved their plasma [Na+] as compared to rats refed protein deficient diets, and this improvement was accompanied by decreases in natriuresis despite an increased glomerular filtration rate and an unchanged negative free water clearance. Identical effects were observed in rats refed a protein deficient diet but supplemented with oral urea, suggesting that urea generation from catabolism of dietary protein is responsible for the effect of protein refeeding to decrease urinary sodium excretion. Both the protein and urea refed rats had significantly higher inner medullary urea contents and concentrations compared to rats refed protein-deficient diets and also to rats studied immediately before protein refeeding, supporting the hypothesis that urea and dietary protein decrease natriuresis in patients with SIAD in association with increased inner medullary urea concentrations.

Localization and characterization of renal calcitonin receptors by in vitro autoradiography.

Calcitonin receptor binding sites were identified in renal cortex and medulla using the radioligand 125I-salmon calcitonin. Microscopic localization of these receptors revealed binding over medullary and cortical thick ascending limb of the loop of Henle and in distal convoluted tubule. A number of receptor positive cells in the inner medulla were also identified. Characterization of the binding demonstrated a single class of high-affinity binding sites in both the medulla and the cortex with affinity constants of 0.74 +/- 0.09 x 10(9) M-1 and 0.32 +/- 0.05 x 10(9) M-1, respectively, and receptor concentrations of 205 +/- 45 fmol/mg protein and 453 +/- 54 fmol/mg protein, respectively. Competition for 125I-salmon calcitonin binding by a wide range of calcitonin analogs revealed a close correspondence between the reported biological potencies and activities in the current system. The localization of binding sites within the nephron corresponds to the reported localization of calcitonin-stimulated adenylate cyclase activity and suggests that the receptor mediated actions of calcitonin in the kidney utilize cyclic AMP as a second messenger. In addition, the microscopic identification of specific calcitonin receptors helps the delineation of direct actions of this hormone from those which are indirect.