Acetazolamide Attenuates Lithium–Induced Nephrogenic Diabetes Insipidus

Abstract

To reduce lithium–induced nephrogenic diabetes insipidus (lithium-NDI), patients with bipolar disorder are treated with thiazide and amiloride, which are thought to induce antidiuresis by a compensatory increase in prourine uptake in proximal tubules. However, thiazides induced antidiuresis and alkalinized the urine in lithium-NDI mice lacking the sodium-chloride cotransporter, suggesting that inhibition of carbonic anhydrases (CAs) confers the beneficial thiazide effect. Therefore, we tested the effect of the CA–specific blocker acetazolamide in lithium-NDI. In collecting duct (mpkCCD) cells, acetazolamide reduced the cellular lithium content and attenuated lithium-induced downregulation of aquaporin-2 through a mechanism different from that of amiloride. Treatment of lithium-NDI mice with acetazolamide or thiazide/amiloride induced similar antidiuresis and increased urine osmolality and aquaporin-2 abundance. Thiazide/amiloride-treated mice showed hyponatremia, hyperkalemia, hypercalcemia, metabolic acidosis, and increased serum lithium concentrations, adverse effects previously observed in patients but not in acetazolamide-treated mice in this study. Furthermore, acetazolamide treatment reduced inulin clearance and cortical expression of sodium/hydrogen exchanger 3 and attenuated the increased expression of urinary PGE2 observed in lithium-NDI mice. These results show that the antidiuresis with acetazolamide was partially caused by a tubular-glomerular feedback response and reduced GFR. The tubular-glomerular feedback response and/or direct effect on collecting duct principal or intercalated cells may underlie the reduced urinary PGE2 levels with acetazolamide, thereby contributing to the attenuation of lithium-NDI. In conclusion, CA activity contributes to lithium-NDI development, and acetazolamide attenuates lithium-NDI development in mice similar to thiazide/amiloride but with fewer adverse effects.

Lithium is the drug of choice for the treatment of bipolar disorders, and it is also regularly used to treat schizoaffective disorders and depression. Lithium is a frequently prescribed drug, because it is provided to 0.1% of the Western population. Unfortunately, in 2%–85% of patients and depending on age, lithium usage leads to nephrogenic diabetes insipidus (NDI), a disorder characterized by an impaired response of the kidney to vasopressin (arginine vasopressin [AVP]), leading to polyuria and polydipsia.^1–3 Patients with lithium–induced nephrogenic diabetes insipidus (Li-NDI) are at risk for dehydration–induced lithium toxicity, and prolonged lithium treatment might lead to cyst formation and ESRD.⁴ However, cessation of lithium therapy is not an option for most patients with NDI, because bipolar disorder symptoms have a larger effect on the patient’s quality of life.

From studies in rats, it became clear that Li-NDI develops in two stages. In the short term (10 days), Li-NDI coincides with downregulation of aquaporin-2 (AQP2) water channels, which is caused by a reduced AQP2 transcription.^5–7 Despite an increased proliferation of the AQP2–expressing principal cells of the collecting duct, long–term lithium treatment (4 weeks) also results in a severe loss of AQP2–expressing principal cells, which might be attributed to a lithium–induced G2/M–phase cell cycle arrest.^8,9 This principal cell loss is compensated for by an increased number of α-intercalated cells, which are involved in acid secretion.⁸

To reduce polyuria in patients receiving lithium, a low-sodium diet together with thiazide and amiloride diuretics are prescribed.¹⁰ Amiloride acts on the principal cell epithelial sodium channel (ENaC), and we and others found that amiloride blocks principal cell lithium entry through ENaC, thereby attenuating polyuria in rodents and humans.^11–13 Thiazides are known to block sodium and chloride reabsorption through the NaCl cotransporter (NCC) in the renal distal convoluted tubule, and the antidiuretic effect has been ascribed to a hypovolemia-induced activation of the renin-angiotensin-aldosterone system and a compensatory increased uptake of sodium and water in proximal tubules. Recently, however, we discovered that thiazide also has an NCC-independent effect, because NCC knockout mice with Li-NDI showed a clear antidiuretic response on treatment with thiazide.¹⁴

Because urine of our thiazide-treated mice was alkalinized and thiazides are derived from carbonic anhydrase (CA) inhibitors,¹⁵ our data indicated that the antidiuretic effect of thiazide in Li-NDI may involve CA inhibition. CAs catalyze the hydration of carbon dioxide to form carbonic acid, which then rapidly dissociates to form protons and bicarbonate, and they play major roles in pH balance regulation. Here, we show that CAs are, indeed, involved in lithium–induced AQP2 downregulation and that, by inducing a tubular glomerular feedback response and through direct action on collecting duct cells, the CA–selective drug acetazolamide not only attenuates Li-NDI but yields superior in vivo effects compared with the presently used treatment for Li-NDI.

Results

The Clinically Used Drug Acetazolamide Attenuates Lithium-Induced Downregulation of AQP2 in mpkCCD Cells

mpkCCD cells are mouse collecting duct cells showing dDAVP-dependent expression of endogenous AQP2, and we have shown that thiazide reduces lithium-induced downregulation of AQP2 in these cells, whereas they lack NCC expression.^7,11,14 Because our previous animal studies suggested that thiazides reduced polyuria in our NCC knockout mice by inhibiting CAs, we wanted to test whether acetazolamide, a stable CA inhibitor that is commonly used in patients, could also rescue lithium–induced AQP2 downregulation in mpkCCD cells. Indeed, whereas lithium again reduced the AQP2 abundance in mpkCCD cells, acetazolamide significantly attenuated this downregulation (Figure 1, A and B). Because our data suggest that both thiazide and acetazolamide influence lithium–reduced AQP2 abundances through CAs, we assessed whether the action mechanism of acetazolamide differs from that of amiloride. If so, we anticipated that acetazolamide and amiloride together should attenuate the lithium–induced AQP2 downregulation better than cells treated with amiloride only. Indeed, immunoblotting revealed a significantly higher AQP2 abundance in cells treated with amiloride and acetazolamide compared with amiloride only (Figure 1, C and D). We and others discovered that ENaC is the main cellular entry site for lithium and that amiloride strongly reduced the intracellular lithium levels in mpkCCD cells.^11,14 Determination of the intracellular lithium concentrations revealed that amiloride, indeed, reduced the intracellular lithium concentration by 87%, whereas this was only 30% with acetazolamide (Figure 1E). The mean intracellular lithium concentration was nominally lower with amiloride/acetazolamide; however, there was no significant difference (P=0.23). Transcellular transport of sodium and potassium through ENaC, renal outer medullary K⁺, and the Na/K-ATPase is electrogenic and therefore, generates a transcellular voltage (Tv) over mpkCCD cell monolayers. Lithium slightly reduced the Tv, which was not further decreased with acetazolamide (Figure 1F). In contrast, amiloride completely blocked the Tv in lithium–treated mpkCCD cells. Together, these data reveal that the CA–specific inhibitor acetazolamide attenuates lithium-induced downregulation of AQP2 in vitro and that its mechanism of action is different from that of amiloride.

Figure 1.

Acetazolamide (Acz) reduces lithium (Li⁺) -induced downregulation of AQP2 abundance in mpkCCD cells. Native mpkCCD cells were grown to confluence for 4 days and subsequently exposed to 1 nM dDAVP for another 4 days. During the last 2 days, cells were incubated in the absence (control [Ctr]) or presence of Li⁺ only or with Li⁺ and 100 μM Acz, 10 μM amiloride (Am), 100 μM hydrochlorothiazide and Am (T+Am), or Am and Acz (Am+Acz). At the basolateral and apical sides, final concentrations of 1 and 10 mM Li⁺ were used, respectively. (F) After measurements of Tv, (A and C) cells were lysed and subjected to AQP2 immunoblotting. Molecular masses (in kilodaltons) are indicated. (B and D) The signals for nonglycosylated (29 kD) and complex-glycosylated (40–45 kD) AQP2 were densitometrically quantified. Mean values±SEMs of normalized AQP2 abundance are given relative to Ctr. (E) Intracellular Li⁺ concentrations were determined, corrected for contamination with extracellular Li⁺, and normalized for the amount of protein ([Li⁺]±SEM in picomoles per microgram protein). Data from three independent experiments (one-way ANOVA and Bonferroni multiple comparison test). *P<0.05.

Acetazolamide Attenuates Development of Li-NDI in Mice

To investigate whether acetazolamide attenuates development of Li-NDI, mice were maintained on lithium chow only or lithium combined with acetazolamide or thiazide/amiloride for 10 days. As reported,^11,16 mice treated with lithium developed severe polyuria and polydipsia combined with a significantly reduced urine osmolality (Figure 2). Interestingly, acetazolamide treatment induced a significant antidiuresis and increase in urine osmolality, which was slightly but not significantly better than that in mice treated with thiazide/amiloride. Consistent with the induced antidiuresis, water intake was significantly reduced with the acetazolamide treatment compared with lithium only.

Figure 2.

Acetazolamide (Acz) attenuates lithium (Li⁺)-induced NDI in mice. (A) Urine volume, (B) water intake, and (C) urine osmolality of untreated mice (control [Ctr]) or mice treated for 10 days with Li⁺ or Li⁺ combined with thiazide/amiloride (T+Am) or Acz. During the last 48 hours, mice were housed in metabolic cages; during the last 24 hours, water intake was measured, and urine was collected to determine urine volume and osmolality (n=8 mice per group; one-way ANOVA and Bonferroni multiple comparison test). *P<0.05.

Because long–term lithium treatment coincides with reduced AQP2 and increased H⁺-ATPase abundance in the kidney,⁶ we also analyzed their abundances. Immunoblot analysis revealed that lithium reduced AQP2 abundance (Figure 3, A and B), which was significantly attenuated by both acetazolamide and thiazide/amiloride (Figure 3, A–D). With acetazolamide or thiazide/amiloride, however, AQP2 levels did not return to control levels. H⁺-ATPase levels were similar for all groups (Supplemental Figure 1), which is in line with the notion that lithium–induced collecting duct remodeling is not present after 10 days of lithium treatment in rodents.^9,16

Figure 3.

Thiazide/amiloride (T+Am) and acetazolamide (Acz) reduce lithium (Li⁺) -induced downregulation of AQP2 in Li-NDI mice. (A–D) Immunoblot and corresponding densitometric analyses of AQP2 of mouse kidneys that are untreated (control [Ctr]), treated with Li⁺ only, or treated with Li⁺ together with Acz or T+Am. (B and D) The signals for AQP2 are densitometrically quantified. Mean values±SEMs of normalized AQP2 abundance are given relative to Ctr. Equal loading of the samples was confirmed by staining of the blots with Coomassie blue (Cm). One-way ANOVA and Bonferroni multiple comparison test. *Significant differences (P<0.05) from Ctr. Paraffin sections of immersion-fixed kidneys from (A, E, and I) Ctr, (B, F, and J) Li⁺-treated, (C, G, and K) Li⁺+T+Am-treated, and (D, H, and L) Li⁺+Acz-treated mice were incubated with a rabbit polyclonal AQP2 antibody followed by a Cy3–coupled goat anti–rabbit IgG. (A–D) Overviews and high magnifications of representative (E–H) connecting tubules (CNTs) and (I–L) cortical collecting ducts (CCDs).

To examine segment-specific effects of the different therapies on AQP2 abundance, immunohistochemistry was performed. Consistent with our immunoblot data, lithium treatment strongly reduced AQP2 staining, which was clearly attenuated in kidneys of mice treated with lithium and acetazolamide or thiazide/amiloride (Figure 3E). Interestingly, although lithium abolished AQP2 expression in the entire kidney, the increased AQP2 abundance in the thiazide/amiloride-treated group mainly localized to the inner medulla of the kidney, whereas in the acetazolamide-treated mice, AQP2 abundance was increased along the connecting tubule and entire collecting duct. Consistent with our immunoblot data, immunohistochemistry revealed no clear changes in H⁺-ATPase labeling between the different groups (Supplemental Figure 2).

Acetazolamide Shows an Improved Overall Electrolyte Balance over Thiazide/Amiloride

Some patients on thiazide/amiloride therapy have been reported to develop hyponatremia, hyperkalemia, metabolic acidosis, and/or hypercalcemia.^17–20 Moreover, initiating thiazide/amiloride treatment in patients with Li-NDI often leads to elevated blood lithium levels, necessitating adjustment of the lithium dose.²¹ Therefore, to assess and compare the effect of acetazolamide and thiazide/amiloride on these parameters, we analyzed blood and urine for lithium and electrolyte levels (Table 1). Indeed, although mice treated with thiazide/amiloride had a reduced body weight and developed hyponatremia, hyperkalemia, and a metabolic acidosis, these parameters were not affected in acetazolamide-treated mice. Also, thiazide/amiloride treatment induced hypercalcemia, which was significantly reduced with acetazolamide, although not to control levels. Moreover and consistent with patients, serum lithium concentrations were significantly increased in our thiazide/amiloride mice but unchanged in our acetazolamide mice compared with lithium controls.

Table 1.

Metabolic parameters of mice treated for 10 days with standard chow only or together with lithium; lithium, thiazide, and amiloride; or lithium and acetazolamide

Metabolic Parameters	Ctr	Li+	Li+ + Am + T	Li+ + Acz
Serum osmolality (mosM/kg)	320±1	319±1	311±0.5a	321±3b
Serum sodium (mmol/L)	150±0.3	149±0.4	139±0.8a	150±0.5b
Serum potassium (mmol/L)	5.3±0.1	5.6±0.2	7.6±0.5a	5.4±0.2b
Serum lithium (mmol/L)	—	0.63±0.04c	2.11±0.12a	0.69±0.04b
Serum creatinine (mg/dl)	0.08±0.01	0.09±0.00	0.06±0.01a	0.04±0.01a,b
Blood ionized calcium (mmol/L)	1.24±0.00	1.32±0.01c	1.34±0.01	1.27±0.01a,b
Blood pH	7.34±0.01	7.32±0.01	7.24±0.02a	7.35±0.02b
Urine sodium (mmol/L)	352±40	46±6c	484±68a	112±12b
Urine potassium (mmol/L)	870±83	131±15c	154±19	271±31a,b
Urine lithium (mmol/L)	—	20±3c	24±3	38±8
Urine creatinine (mg/dl)	70±7	10±1c	14±2	18±2a
Total sodium excretion (mmol/24 h)	0.17±0.02	0.16±0.01	1.26±0.12a	0.22±0.02b
Total potassium excretion (mmol/24 h)	0.42±0.05	0.55±0.02c	0.40±0.03a	0.54±0.05
Total lithium excretion (mmol/24 h)	—	84±3c	64±6	72±11
Body weight (g)	18.9±0.4	18.0±0.2c	16.0±0.5a	17.4±0.4b
Food intake (mg/g per 24 h)	214±8	198±10	188±5	228±5a,b
Feces production (mg/g per 24 h)	111±7	98±8	82±2	123±9b

Values are means±SEMs. Ctr, control (standard chow only); Li⁺, lithium; Am, amiloride; T, thiazide; Acz, acetazolamide; —, below detection limit.

^aP<0.05 compared with lithium treatment.

^bP<0.05 compared with lithium, thiazide, and amiloride treatment.

^cP<0.05 compared with control treatment.

The Antidiuretic Effect of Acetazolamide Coincides with a Lowered GFR and a Reduced Prostaglandin E2 Release

Paradoxically, the clear antidiuresis found in our acetazolamide group (Figure 2) coincided with a significantly increased creatinine clearance (Figure 4A). However, because the creatinine clearance (especially in mice) highly depends on both creatinine secretion and reabsorption in the proximal tubule, the segment mainly influenced by acetazolamide,^22,23 we hypothesized that acetazolamide may affect proximal tubular creatinine reabsorption/secretion in our mice and thus, that the creatinine clearance did not properly reflect the GFR. Therefore, we used FITC-inulin to determine the GFR in an identically performed animal experiment. Although acetazolamide again significantly attenuated Li-NDI (not shown), the clearance of FITC-inulin was significantly reduced with acetazolamide (Figure 4B), indicating that acetazolamide reduced the GFR. Urinary prostaglandin E2 (PGE2), which extensively contributes to AQP2 downregulation in Li-NDI,²⁴ was significantly increased in our Li-NDI mice but fully attenuated in our mice treated with lithium/acetazolamide (Figure 4C). Moreover and consistent with its CA inhibitory action in proximal tubules, acetazolamide further increased urinary pH (Figure 4D) and strongly reduced the abundance of NHE3 in the renal cortex compared the cortex of mice treated with lithium only (Figure 4, E and F).

Figure 4.

Acetazolamide (Acz) reduces the GFR and abolishes the elevated PGE2 levels in lithium-treated mice. Mice were treated for 10 days with control (Ctr) diet or diet containing lithium (Li⁺) only or Li⁺ combined with Acz. During the last 48 hours, mice were housed in metabolic cages, and during the last 24 hours, urine was collected to determine (A) urinary pH, (D) creatinine clearance, and (F) PGE2 levels. (C) At day 10, mice were euthanized, and blood and kidneys were isolated, enabling the analysis of renal NHE3 abundance. In B, the arrow indicates the approximately 85-kD band of NHE3. (E) To measure GFR using FITC-inulin, the above-mentioned experiment was repeated; however, at day 4, osmotic minipumps containing FITC-inulin were implanted, and at day 10, FITC-inulin levels were measured in 24-hour urine and serum (n=8 mice per group). One-way ANOVA and Bonferroni multiple comparison test. Cm, Coomassie blue. *P<0.05.

Discussion

Acetazolamide Is Superior to Thiazide/Amiloride to Attenuate Li-NDI

Our mouse studies revealed that acetazolamide attenuates development of Li-NDI to the same extent as thiazide/amiloride, but in contrast to acetazolamide, our thiazide/amiloride-treated mice developed hyponatremia, hyperkalemia, and metabolic acidosis. In humans, hyponatremia is mostly a consequence of upregulated AQP2 expression by high circulating AVP levels. Although AVP levels are elevated in Li-NDI, this cannot explain hyponatremia in our thiazide/amiloride-treated mice, because AQP2 is downregulated here. Instead, our data indicate that the hyponatremia is because of the induced natriuresis caused by thiazide and amiloride in a status of polyuria and polydipsia, because our Li-NDI mice were normonatremic and our thiazide/amiloride mice were highly natriuretic compared with the other groups. Note, however, that part of the increased natriuresis must be because of an increased consumption of salt from the provided salt block, because food intake was not increased. The mice apparently drank water to satiety, because the hematocrit was not different between the groups (data not shown). Similarly, patients with congenital NDI also sometimes develop hyponatremia when treated with thiazide combinations.^25,26

The observed hyperkalemia is likely caused by inhibition of ENaC by amiloride, because renal secretion of potassium occurs only in exchange of ENaC–mediated sodium reabsorption.²⁷ Lithium itself can lead to metabolic acidosis,²⁸ which would even be increased with inhibition of bicarbonate uptake by acetazolamide, but we only found metabolic acidosis in our thiazide/amiloride-treated mice. Whereas metabolic acidosis in our acetazolamide group may have been compensated for by increased ventilation, metabolic acidosis in the thiazide/amiloride group may be secondary to the observed hyperkalemia, because mammals attenuate hyperkalemia at the expense of development of metabolic acidosis.²⁹

Our Li-NDI mice developed hypercalcemia, which was sustained in our thiazide/amiloride-treated mice but not in our acetazolamide mice. Hyperparathyroidism is common with lithium-using patients, and the occurrence of hypercalcemia in Li-NDI has been ascribed to inhibition of calcium–sensing receptor signaling by lithium in the parathyroid.^30–32 The corrected blood calcium levels with acetazolamide, however, may be unrelated to the parathyroid effect of lithium, because plasma calcium is increased by bone resorption, a process that involves CA2 activity in osteoclast and is strongly inhibited by acetazolamide.^33,34

Another important advantage of the use of acetazolamide over thiazide/amiloride is that plasma lithium concentrations remained unchanged with acetazolamide. Blood lithium is mainly set by the amount reabsorbed in proximal tubules, a process in which the apical NHE3 is highly involved, and it is stimulated by thiazide.^35,36 Considering this, the unaltered blood lithium levels with acetazolamide are best explained by combinatory effects of the reduced NHE3 abundance in proximal tubules and the reduced GFR.

Acetazolamide Attenuates Li-NDI by a Dual Mode of Action

Our data indicate that the observed antidiuresis and reduced GFR with acetazolamide is because of a tubular glomerular feedback response caused by inhibition of CAs in the proximal tubule.^37,38 Ninety percent of renal HCO₃⁻ is reabsorbed in proximal tubules, which is strongly facilitated by CA4/14 (luminal/apical), CA2 (intracellular), and CA4/12 (basolateral) hydrating CO₂ and dehydrating H₂CO₃.³⁹ In this process, secretion of protons by NHE3 is important. By blocking these CAs, acetazolamide prevents the intracellular generation of H⁺, which is needed for NHE3 to reabsorb filtered Na⁺.^40–43 The consequently increased tubular salt and water load in the proximal tubule leads to an increased fluid delivery and tubular [Cl⁻] at the macula densa, which induces a tubular glomerular feedback response (i.e., a reduced GFR).³⁸ Indeed, acetazolamide in our Li-NDI mice led to an increased urinary pH, reduced NHE3 abundance, and reduced GFR, which is in agreement with reported data on NHE3 knockout mice.^38,41,44

Other than intercalated cells (see below), the increased fluid delivery to the macula densa with acetazolamide may also partially explain the observed lower urinary levels of PGE2, thereby attenuating Li-NDI. By acting on EP1/3 receptors, increased urinary PGE2 levels in Li-NDI reduce principal cell AQP2 expression and thus, water reabsorption.⁴⁵ In Li-NDI, a fraction of the elevated urinary PGE2 levels is thought to be derived from macula densa and surrounding cTAL cells, which produce PGE2 to increase renin synthesis and release in response to a reduced fluid delivery to the TAL/hypovolemia.^43,44 As such, the increased fluid delivery to the TAL with acetazolamide will reduce the cortical release of PGE2 and therefore, Li-NDI.

However, our in vitro data indicate that acetazolamide also directly protects collecting duct cells from lithium, but it is at present unclear whether in vivo acetazolamide acts directly on principal cells or indirectly through intercalated cells. Support for the first is that mpkCCD cells endogenously express and show proper regulation of the typical principal cell proteins AQP2 and ENaC. Moreover, mpkCCD cells express high CA2 mRNA levels (http://esbl.nhlbi.nih.gov/mpkCCD-transcriptome/), which are also expressed in principal cells in vivo.^39,46–48 Also, the in vivo activity of ENaC, the lithium entry site of principal cells, has been reported to be functionally paired with CA activity, because CA inhibition by acetazolamide reduced the intracellular pH and ENaC activity in sweat duct cells and colon.^49,50

However, mpkCCD cells may not fully represent principal cells, and because intercalated cells express abundant levels of CA2, -4, -12, and -15,³⁹ acetazolamide may increase principal cell AQP2 expression and water uptake indirectly by inhibiting ACs in intercalated cells. Indeed, long–term lithium treatment leads to metabolic acidosis, which underlies the increased number of α-intercalated cells,^11,51 and because lithium inhibits intercalated cell H⁺-ATPase and H⁺/K⁺-ATPase activity,^52,53 it has been suggested that acidosis-induced proliferation of α-intercalated cells may contribute to Li-NDI.¹⁶ It is unlikely, however, that attenuation of Li-NDI in our mice is caused by direct action of acetazolamide on α-intercalated cells or collecting duct remodeling for several reasons. First, acetazolamide increases the number of α-intercalated cells in rodents, because it causes acidosis itself.^54,55 Second and consistent with the unchanged H⁺-ATPase expression in our mice, collecting duct remodeling is not observed within 10 days of lithium treatment but only starts at about 4 weeks of treatment.^9,56

An effect of acetazolamide on β-intercalated cells, however, may be more likely. Although lithium treatment did not reduce the number of these cells,⁵⁷ exciting recent studies revealed the existence of extensive cross-talk between β-intercalated and principal cells in the regulating collecting duct function.^58–61 Although Eladari and coworkers⁶² elegantly showed that, in rodents, the sodium–dependent chloride bicarbonate exchanger (NDCBE; SLC4a8) and chloride bicarbonate exchanger protein pendrin allow for NaCl reabsorption through β-intercalated cells, chloride permeation through pendrin also seemed necessary for ENaC–mediated sodium reabsorption and expression. Also, in mice lacking functional intercalated cell–specific H⁺-ATPase, the observed natriuresis and aquaresis were caused by dysfunctional and lower abundances of ENaC and pendrin/NDCBE and reduced AQP2 levels, respectively.⁶⁰ Eladari and coworkers⁶² further showed that the lack/inhibition of β-intercalated cells H⁺-ATPase led, through flow–stimulated luminal ATP release, to autocrine and paracrine release of PGE2, which reduced the cortical and medullary ENaC activity and AQP2 abundance. Importantly, thiazides inhibited NCC–independent NaCl reabsorption through NDCBE/pendrin,⁵⁹ and acetazolamide reduced pendrin abundance,⁶³ drugs that we showed to attenuate Li-induced downregulation of AQP2 and Li-NDI through a similar mechanism. As such, the attenuated Li-NDI with acetazolamide, which was given during the entire lithium treatment, may be caused by an impaired functioning of pendrin, resulting in increased ATP/PGE2 release, reduced ENaC activity in principal cells, and thus, reduced influx of lithium from prourine. The finding that acetazolamide is beneficial chiefly in the cortical segments that contain intercalated cells is consistent with the possibility that intercalated cell CAs could be involved. A prime candidate here is CA12, because it is highly sensitive to acetazolamide, and patients with reduced CA12 activity have a preponderance to hyponatremic dehydration.⁶⁴ Whether one of these mechanisms underlies the beneficial effect of acetazolamide in Li-NDI remains to be studied.

Taken together, we have shown that CA activity contributes to Li-NDI development, that acetazolamide attenuates Li-NDI by inducing a tubular glomerular feedback response and through a direct action on the collecting duct, and that acetazolamide attenuates Li-NDI development similar to thiazide/amiloride but with fewer side effects.

Concise Methods

Cell Culture

mpkCCD cells were cultured as described.⁶⁵ Cells were seeded at a density of 1.5×10⁵ cells per centimeter² on semipermeable filters (Transwell; 0.4-μm pore size; Corning Costar, Cambridge, MA) and cultured for 8 days. Unless stated otherwise, the cells were exposed to 1 nM dDAVP at the basolateral side for the last 96 hours to induce AQP2 expression. Lithium and compounds were administered as indicated. At the end of the experiment, transcellular electrical resistance and voltage were measured using a Millicell-ERS Meter (Millipore Corp., Bedford, MA). On day 8, cells were harvested and lysed in Laemmli buffer for Western blotting or stored in Trizol Reagent (Invitrogen, Carlsbad, CA) at −80°C for RNA isolation.

Lithium Assays

Determination of intracellular lithium concentrations was done as described.¹¹ Shortly, mpkCCD_cl4 cells were grown on 4.7-cm² filters. To determine the extent of lithium contamination from the extracellular side, FITC-dextran was added to the lithium-containing medium to a final concentration of 10 μM just before harvesting, after which the medium was mixed. Then, the filters were washed three times with iso-osmotic sucrose (pH 7.3) at 4°C, and cells were lysed by sonication in 1 ml milli-Q Water. Of the 800-μl sample, the amount of lithium was determined by flame photometry, from which the total amount of lithium in the sample was calculated.

Of the 100-μl sample, the amount of FITC-dextran was measured using spectrofluorophotometry (RF-5301; Shimadzu, Tokyo, Japan) at 492-nm (excitation) and 518-nm (emission) wavelengths. By comparing the obtained values with a 2-fold FITC-dextran dilution series, the FITC-dextran concentration in each sample was determined, from which the extent of extracellular lithium contamination was calculated. This was subtracted from the total amount to obtain the intracellular lithium amount. With the used FITC-dextran concentration, a contamination >1:5000 would be detected. To correct for differences in cellular yield, the intracellular lithium amounts were normalized for the protein amount in each sample, which was determined using the Bio-Rad Protein Assay (Bio-Rad, Munich, Germany).

Experimental Animals

Eight- to 10-week-old female C57Bl6/JOlaHsd mice (Harlan Laboratories Inc., Frederick, MD) were maintained in a temperature-controlled room with lights on from 8:00 AM to 8:00 PM. They received normal diet (ssniff R/M-H V1534; ssniff Spezialdiaten GmbH, Soest, Germany) with additions (see below) and water ad libitum for 10 days. For the experiments, mice were divided into four groups (n=8), which were treated as follows: group 1: control mice given a normal diet; group 2: normal diet with 40 mmol LiCl per 1 kg dry food⁶⁶; group 3: diet of group 2 with 200 mg amiloride¹¹ and 350 mg hydrochlorothiazide per 1 kg dry food³⁶; and group 4: diet of group 2 with 180 mg acetazolamide per 1 kg dry food.⁶⁷ LiCl, amiloride, hydrochlorothiazide, and acetazolamide were solubilized in water and mixed with the chow, after which it was dried. All mice had free access to water, food, and a sodium-chloride block.

For the last 48 hours of the experiment, mice were housed in metabolic cages to measure water intake and urine output during the last 24 hours. Mice were anesthetized with isofluorothane, after which their blood was removed by orbita extraction. Then, mice were killed by cervical dislocation, and the kidneys were rapidly removed. One kidney was processed for immunohistochemistry, whereas the other kidney was used for immunoblotting, both as described below. For immunoblotting, the tissue was homogenized using a Polytron Homogenizer (VWR International, Amsterdam, The Netherlands) in 1 ml ice–cold homogenization buffer A (20 mM Tris, 5 mM MgCl₂, 5 mM Na₂HPO₄, 1 mM EDTA, 80 mM sucrose, and protease inhibitors [1 mM PMSF, 5 µg/ml pepstatin A, 5 µg/ml leupeptin, and 5 µg/ml a-protinin]), cleared from nuclei and unbroken cells by centrifugation at 4000×g for 15 minutes, and diluted in Laemmli buffer to a final protein concentration of 1 μg/μl.

Determination of GFR Using FITC-Inulin

To determine the GFR by the FITC-inulin clearance method,^68,69 we used mice as described above, and these mice were also treated as before (n=8 per group). Four days after the start of the diet, minipumps (Model 2001; Alzet, Cupertino, CA) containing 3% FITC-inulin were subcutaneously implanted in the isofluorane-anesthetized mice. At treatment days 9 and 10, mice were housed in metabolic cages, and 24-hour urine was collected in amber tubes at day 10. During this 24 hours, metabolic cages and urine collection tubes were covered with aluminum foil to prevent exposure to light. Traces of left FITC-inulin urine in metabolic cages were added to the collected urine by washing the cage with 5 ml 500 mM HEPES buffer. On day 10, mice were anesthetized with isofluorane, blood was collected by retro-orbital bleeding, and mice were killed by cervical dislocation. Urine fluorescence was determined using a Cytofluor II Fluorescence Multiwell Plate Reader (PerSeptive Biosystems, Framingham, MA) with 485-nm excitation and 538-nm emission. The excretion rate of inulin (24-hour urinary fluorescence counts/plasma fluorescence counts per ml) was taken as the GFR.

Blood and Urine Analyses

Whole blood was analyzed immediately for sodium, potassium, hematocrit, and pH using the EG7+ Cartridge and the I-Stat Clinical Analyzer (Abbott BV, Hoofddorp, The Netherlands). The remaining blood was collected in a BD Microtainer SST Tube (REF 365968; Becton Dickinson BV, Breda, The Netherlands) for serum and centrifuged at 10,000×g for 3 minutes to sediment the red blood cells. Serum and urine samples were analyzed for osmolality using an osmometer (Fiske, Needham Heights, MA), and electrolyte concentrations were measured on a Synchron CX5 Analyzer (Beckman Coulter, Inc., Brea, CA) following the manufacturer’s protocols. Urine PGE2 levels were determined by measuring stable prostaglandin E2 metabolite (PGEM) after chemical derivation of PGE2 and its primary metabolites 13,14-dihydro-15-keto PGE2 and 13,14-dihydro-15-keto PGA2 to the single PGEM compound. PGEM concentrations were determined with the Prostaglandin E Metabolite EIA Kit (Cayman Chemicals, Ann Arbor, MI) according to the manufacturer’s instructions.

Immunoblotting

mpkCCD cells from 1.13-cm² filters were lysed in 200 μl Laemmli buffer and sonicated. mpkCCD lysate and 5–10 μg kidney material in Laemmli were denatured for 30 minutes at 37°C. Protein concentration was determined using the Bio-Rad Protein Assay (Bio-Rad) according to manufacturer’s instructions. SDS-PAGE, blotting, and blocking of the PVDF membranes were done as described.⁷⁰ Membranes were incubated for 16 hours at 4°C with 1:2000 affinity–purified rabbit pre–C tail AQP2 antibody recognizing amino acids 236–255⁷¹ in Tris-Buffered Saline Tween-20 supplemented with 1% nonfat dried milk. In an identical way, other blots were incubated with a rabbit CA12 antibody (gift from William S. Sly, St. Louis University School of Medicine, St. Louis, MO) and a rabbit CA2 antibody (Abcam, Inc., Cambridge, MA). After washing in Tris-Buffered Saline Tween-20, all blots were incubated for 1 hour with 1:5000-diluted goat anti–rabbit IgGs (Sigma-Aldrich, St. Louis, MO) as secondary antibody coupled to horseradish peroxidase. Proteins were visualized using enhanced chemiluminescence (ECL; Pierce, Rockford, IL). Densitrometric analyses were performed using Bio-Rad quantification equipment (Bio-Rad 690c Densitometer; Chemidoc XRS) and software (QuantityOne; Bio-Rad). Equal loading of the samples was confirmed by staining of the blots with Coomassie blue.

Immunohistochemistry

Kidneys were fixed by immersion for 24 hours in 4% paraformaldehyde in 0.1M phosphate buffer at 4°C, embedded in paraffin, and cut into 3- to 4-μm-thick sections. After deparaffinization, sections were placed into a microwave oven and heated for 10 minutes at 98°C in 0.01 M sodium-citrate buffer (pH 6.0) for antigen retrieval. Subsequently, sections were incubated overnight at 4°C with 1:80,000–diluted rabbit polyclonal AQP2 antibodies or 1:2000–diluted rabbit polyclonal H⁺-ATPase antibodies as described.^72,73 The bound primary antibodies were revealed with Cy3–coupled goat anti–rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). To check for unspecific binding of primary or secondary antibodies, incubations with nonimmune sera or without any primary antibodies were performed. All control experiments were negative. Cryosections were studied by epifluorescence using a Leica Microscope (Leica Microsystems, Buffalo Grove, IL). Connecting tubules and cortical collecting ducts were distinguished on the basis of their specific localization in the cortical labyrinth and the medullary rays, respectively. Images were acquired with a charge–coupled device camera. For overviews, single images were taken with the automated scanning mode of the microscope and afterward, stitched using the Leica Application Suite. Digital images were processed electronically with Adobe Photoshop (Adobe Systems, Inc., San Jose, CA) and Microsoft Powerpoint (Microsoft, Redmond, WA) software. Adjustments for brightness and contrast were kept constant for each kidney section.

Statistical Analyses

One-way ANOVA with Bonferroni correction was applied. A P value of <0.05 was considered significant. Data are presented as means and SEMs.

Study Approval

All animal studies (DEC no. 2011–010) were approved by the Animal Ethical Committee of the Radboud University Medical Center.

Disclosures

None.