Cerebral influx of Na⁺ and Cl⁻ as the osmotherapy-mediated rebound response in rats

This study was performed in accordance with the European Community guidelines for the use of experimental animals using protocols approved either by the Lower Saxony State Office for Consumer Protection and Food Safety (Niedersachsen, Germany) or Supervisory Authority on Animal Testing (Danish Veterinary and Food Administration, Denmark). To avoid variation due to mixed gender, only female Sprague–Dawley rats were employed, aged 9–13 weeks (Taconic A/S, Lille Skensved, Denmark or Janvier Labs, Le Genest-Saint-Isle, France). Whether the present findings hold for male rats as well will require further studies in the future. Rats were housed in groups of 2–5 per cage (Tp III cages, 22 °C, 12:12 h light/dark cycle) with access to unlimited water and standard altromin rodent diet. The allocation of rats into the treatment groups was randomized, and all experiments were reported in compliance with the ARRIVE guidelines (Animal Research: Reporting in Vivo Experiments) [13].

Rats were anesthetized using isoflurane inhalation mixed in O₂ (1.5–5%, 1 l/min) and anaesthesia was maintained throughout the entire experiment. The body temperature was controlled to 37 °C using an electric heating pad (Harvard Apparatus, Holliston, MA, US) and monitored by a rectal probe during the entire procedure. To avoid systemic regulation of blood osmolytes upon hyperosmotic treatment, a functional nephrectomy was performed immediately prior to the initiation of the experiment in all animals except for naïve animals, which were not exposed to isosmolar- or hyperosmolar treatment but underwent anaesthesia induction shortly before decapitation, see Table 1 for grouping of experimental animals. In brief, laparotomy incision areas were treated with local analgesia [2–4 drops 2% tetracaine (Sigma-Aldrich, Brøndbyvester, Denmark, T7508) or xylocaine (1 mg/ml, AstraZeneca A/S, Copenhagen, Denmark, N01BB02)/bupivacaine (0.5 mg/ml, Amgros I/S, Copenhagen, Denmark, N01BB01) (both in 0.9% w/v NaCl)] prior to opening of the abdominal cavity either from the dorsal or the ventral side in the fully anesthetized animals. The renal artery and vein were ligated using non-absorbable suture. For rats given ventral incision, a catheter (for i.p. delivery, see below) was placed during suturing of the incision, while for rats with dorsal incisions, the smaller openings were closed with metal wound clamps immediately after i.p. delivery. The rats received a single i.p. bolus of a physiological NaCl solution (0.9% w/v NaCl) as an isosmolar control treatment, while an equiosmolar bolus of NaCl (1.17 g/kg, 1 M [14]) or mannitol (7.29 g/kg, dissolved in 0.9% w/v NaCl; 2 M) was given to elevate the plasma osmolarity to a similar extent. All solutions were heated to 37 °C and delivered as 2 ml/100 g body weight. We employed i.p. delivery of the osmotic agent as this delivery route gives similar plasma osmolarities as i.v. delivery [14]. For i.v. inhibitor experiments, a catheter was inserted into the tail vein and an inhibitor mixture containing bumetanide (10 mg/kg [11], Sigma-Aldrich, B3023), amiloride (6 mg/kg [15], Sigma-Aldrich, A7410), and methazolamide (20 mg/kg [16], Sigma-Aldrich, SML0720) or vehicle (specified below) was injected 5 min prior to i.p. treatment with isosmolar- or hyperosmolar NaCl, see Table 1 for grouping of experimental animals. Inhibitors were given in a mixture to minimize the number of rats used for experiments. While drug concentrations in the blood are difficult to assess due to unspecific binding to tissue and blood proteins, we estimate maximal blood concentrations of 0.4 mM for bumetanide and amiloride, and 1.2 mM for methazolamide based on estimated blood volume of 7% of the rat body weight (average: 233 g). In a few experiments, rats were given a triple inhibitor or vehicle dose into the tail vein. In this case, a bolus injection of inhibitors or vehicle was given 20 min and 5 min before and 15 min after delivery of isosmolar or hyperosmolar NaCl. In other experiments, rats were positioned in a stereotactic frame (Stoelting, Wood Dale, IL, US, 51500) and a micro drill (CircuitMedic, Haverhill, MA, US, 110-4102) employed to induce a burr hole in the skull (coordinates from bregma: 1.4 mm lateral, 0.8 mm posterior). A Hamilton syringe (G27, Agntho’s AB, Lidingö, Sweden, 2100521) filled with inhibitor mixture (bumetanide: 33 μM, amiloride and methazolamide: 167 μM, final ventricular concentrations estimated to be 20 and 100 μM [17‐22]) or vehicle dissolved in equilibrated (95% O₂/5% CO₂) artificial CSF (aCSF) (120 mM NaCl, 2.5 mM KCl, 1.3 mM MgSO₄ × 7 H₂O, 1 mM NaH₂PO₄, 25 mM NaHCO₃, 10 mM glucose × H₂O, 2.5 mM CaCl₂, pH 7.4 at 37 °C) was fastened to the stereotactic apparatus and introduced into the right lateral ventricle (4.7 mm ventral). Two min prior to isosmolar or hyperosmolar i.p. treatment, 6 μl inhibitor or vehicle solution was injected in 2 s (volume and rate adjusted to hit both lateral ventricles), see Table 1 for grouping of experimental animals. To maintain an optimal intraventricular inhibitor dose, inhibitor or vehicle solution was injected into the ventricular system every 15 min. All the experiments were terminated by decapitation of the animal 1 h after i.p. injection of osmotic agent or physiological saline. A 1 h treatment period was chosen according to the reported near stabilization of plasma osmolarity and brain volume within 30 min after a hyperosmolar challenge [14]. All inhibitor solutions were made freshly each day (some from frozen stock solutions). Bumetanide and methazolamide were dissolved in 0.1 M NaOH (pH adjusted with 0.1 M HCl to pH 11 and 9, respectively) and diluted to 10 mg/ml for injection into the tail vein, while amiloride was dissolved in heated water at 10 mg/ml. Inhibitors, which were introduced into the ventricular system, were dissolved in DMSO (final concentration of 0.2% in aCSF).

The brain was removed immediately after decapitation. The olfactory bulbs and medulla oblongata were discarded and the remaining brain tissue was placed in a pre-weighed porcelain evaporation beaker and weighed within minutes after isolation to reduce loss of brain water. Brain tissue was homogenized in the pre-weighed evaporation beaker using a steel pestle and dried at 100 °C for 3–4 days to a constant mass for determination of the brain water content. The dried brain tissue (75–130 mg) was extracted in 1 ml 0.75 M HNO₃ on a horizontal shaker table for 3 days at room temperature (RT). The Cl⁻ content in the brain extracts was quantified by a colorimetric method using a QuantiChrom™ Chloride Assay Kit (MEDIBENA Life Science & Diagnostic Solution, Vienna, Austria), while the Na⁺ and K⁺ content was quantified using flame photometry (Instrument Laboratory 943, Bedford, MA, US).

A heparin-coated tube (Jørgen Kruuse A/S, Langeskov, Denmark) was filled with pooled blood (venous and arterial) from the neck region upon decapitation of the rats. Blood samples were kept cold for maximal 4 h until centrifugation at 1300g for 10 min at RT. The plasma layer was collected and stored at − 20 °C. The plasma osmolarity was determined by a freezing point depression osmometer (Löser, Berlin, Germany), while the content of Na⁺, Cl⁻, urea and creatinine was measured using a RAPIDLab^® blood gas analyzer (Siemens, Münich, Germany) or flame photometer (Instrument Laboratory 943).

If assuming that the barriers between blood and brain behave as semipermeable membranes, i.e. permeable only to water but not to solutes, a new steady state in brain water content V_h (h; hyperosmolar, in ml/g dry weight) mediated by an elevated plasma osmolarity \({\text{C}}_{\text{osm}}^{\text{h}}\) (mOsm) can be given by Eq. 1 as described in [14], where V_i is brain water content (i; isosmolar, in ml/g dry weight) in rats with isosmolar plasma osmolarity \({\text{C}}_{\text{osm}}^{\text{i}}\) (mOsm).

If the brain water loss is less than predicted by Eq. 1, this will imply that the brain gains osmotically active solutes given that the plasma and brain water is in osmotic equilibrium. The predicted gain of electrolytes, ΔQ (mmol/kg dry weight), can then be given by

To assess the paracellular permeability of the brain barriers, anaesthetized rats were subjected to a functional nephrectomy. The experiments were initiated as above, after which a 4% Na⁺-fluorescein (Sigma-Aldrich, F63772) solution (2 ml/kg, 0.25 ml/min, dissolved in 0.9% w/v NaCl) was infused into the femoral vein through a catheter; Na⁺-fluorescein is a marker of paracellular permeability and has been used to identify paracellular BBB disruption by osmotic shock [23]. Five min hereafter, isosmolar or hyperosmolar NaCl was injected into the abdominal cavity as described above, see Table 1 for grouping of experimental animals. Rats were decapitated after 1 h, and the brains were removed immediately and frozen on crushed solid CO₂. Coronal sections (12 μm) were cut in a cryostat and mounted on slides. Na⁺-fluorescein was visualized using an Axioplan 2 epifluoresence microscope (Carl Zeiss Vision, München-Halbergmoos, Germany) equipped with a Plan Neofluar and an AxioCam MR digital camera by use of the AxioVision 4.4 software (Carl Zeiss Vision, Birkerød, Denmark). Image acquisition was performed in a blinded fashion. Representative images were captured of brain regions comprising the neocortex, hippocampus, thalamus, and the lateral ventricle. The pineal gland was used as an internal positive control due to its lack of BBB [24]. Phase contrast images were included to visualize brain structures in transmitted light. Image processing (brightness and contrast) was performed using Adobe Photoshop (San Jose, CA, US).

In order to monitor the ICP of the anaesthetized rats, a micro drill (1 mm bit) was applied to manually induce a burr hole into the skull until transparency was observed. The thin skull layer was gently ruptured using a 0.6 mm bit (without disruption of dura mater) after which a tweezer was employed to remove skull flakes. An epidural probe (Plastics One, Roanoke, VA, US, C313GS-5-3UP, 0 mm below pedestal) was gently placed onto dura mater, and fastened to the skull by cement (GC, Kortrijk, Belgium, Fuji I, 000136). ICP fluctuations were detected by PicoLog Recorder software (Pico Technology, Cambridgeshire, UK). To ensure proper probe insertion in the epidural space, the jugular vein was compressed before the beginning of each experiment and a raised ICP detected as a positive control. An Evans blue (Sigma-Aldrich, E2129) solution (0.003% w/v in 0.9% w/v NaCl) was infused into the right lateral ventricle (6 μl in total, 3 μl/s) using a Hamilton syringe, while ICP recordings were collected, see Table 1 for grouping of experimental animals. 10 min after intraventricular injections, rats were euthanized by decapitation. The brains were isolated and cerebral hemispheres separated to confirm intraventricular Evans blue staining.

Female Sprague–Dawley rats (22–28 weeks) were anaesthetized with chloral hydrate (400 mg/kg, i.p.), see Table 1 for grouping of experimental animals. A catheter was inserted into the left femoral artery to measure the intra-arterial blood pressure (BioSys software, TSE Systems, Bad Homburg, Germany). The intra-arterial blood pressure was monitored until 1 h after i.v. injection of inhibitors (10 mg/kg bumetanide, 6 mg/kg amiloride and 20 mg/kg methazolamide) or vehicle, and experiments were terminated by decapitation of the rats.

All data are given as mean values ± standard error of mean (SEM). To evaluate statistically significant differences between mean values of two groups, an unpaired two-tailed Student’s t-test was applied, while a one-way analysis of variance (ANOVA) followed by Dunnett’s or Tukey’s multiple comparisons post hoc test was applied to compare mean values of multiple groups. Comparison of two factors was evaluated by a two-way ANOVA followed by Tukey’s multiple comparisons post hoc test. p < 0.05 was considered statistically significant. All statistical analyses were performed in GraphPad Prism 7.0 (GraphPad Software, Inc., La Jolla, CA, US) and indicated in the respective figure legend.

To determine the effect of osmotherapy on the brain water and electrolyte content, we employed a rat in vivo model in which the plasma osmolarity was elevated by i.p. injection of NaCl (1.17 g/kg, 2 ml/100 g body weight). To isolate the effect of brain volume regulation, rats were functionally nephrectomized prior to the procedure, the success of which was evident from the increased plasma content of creatinine and urea in these animals compared to naïve rats, which had not undergone nephrectomy (Fig. 1a, b, see figure legend for values). The plasma osmolarity in the nephrectomized rats treated with isosmolar NaCl (303 ± 1 mOsm, n = 9, termed ‘control’ henceforward) was not significantly different from that of the naïve rats (298 ± 1 mOsm, n = 3, Fig. 1c), indicating that the extended experimental protocol in itself did not interfere with plasma osmolarity. Following a single bolus injection with hyperosmotic NaCl (termed ‘osmotherapy’ henceforward), the plasma osmolarity was increased to 355 ± 1 mOsm after 1 h (n = 9, p < 0.001, Fig. 1c), with an associated increase in the plasma content of Na⁺ and Cl⁻ (n = 9, p < 0.001, Fig. 1d, e, see figure legend for values).

The brain water content of the naïve rats, which were not exposed to isosmolar or hyperosmolar treatment, (3.72 ± 0.03 ml/g dry weight, n = 3) was slightly lower than that of the control rats exposed to the isosmolar NaCl treatment (3.79 ± 0.01 ml/g dry weight, n = 9, p < 0.05, Fig. 2a), while osmotherapy caused a 9% reduction in the brain water content (to 3.46 ± 0.01, n = 9, p < 0.001, Fig. 2a). However, this reduction in brain water content amounted to only ~ 60% of that predicted from ideal osmotic behavior (calculated according to Eq. 1 and illustrated as a dashed red line in Fig. 2a), which indicates that volume regulation takes place. The osmotherapy-mediated reduction in the brain water loss was associated with an increase in brain electrolyte content, with a 15% increase in brain Na⁺ (p < 0.001, Fig. 2b) and a 31% increase in brain Cl⁻ (p < 0.001, Fig. 2c) (see figure legend for values). There was a minor 2% increase in the brain K⁺ content (control: 463 ± 2 mmol/kg dry weight vs. osmotherapy: 471 ± 2 mmol/kg dry weight, n = 9, p < 0.05). The brain Na⁺ and Cl⁻ content in control rats was not significantly different from that obtained in naïve rats, Fig. 2b, c, see figure legend for values. The total increase in osmolyte content represented by Na⁺, Cl⁻, and K⁺, ΔQ_observed, amounted to 79 mmol/kg dry weight, which represents 104% of the predicted osmolyte gain, ΔQ_predicted = 76 mmol/kg dry weight (Eq. 2). The osmotherapy-mediated gain of brain Na⁺ and Cl⁻, and to a minor extent K⁺, can thereby account for the reduction in brain water loss observed 1 h after administration of the hyperosmolar challenge.

To determine the potency of osmotherapy conducted with NaCl vs. mannitol, we performed a parallel experimental series with mannitol as the osmotic agent. The increased Na⁺ and Cl⁻ plasma concentration observed with NaCl infusion (as above, p < 0.001), was absent, and even slightly reversed compared to control rats upon i.p. delivery of mannitol (7.29 g/kg, 2 ml/100 g body weight) (p < 0.001 for Na⁺ and p < 0.05 for Cl⁻, Fig. 2d, e, see figure legend for values). Mannitol treatment yielded a plasma osmolarity (356 ± 3 mOsm, n = 6) similar to that obtained in rats treated with NaCl (355 ± 1 mOsm, n = 9, Fig. 1c, p = 0.71). Mannitol efficiently reduced the brain water content (to 3.51 ± 0.02 ml/g dry weight, p < 0.001), although slightly less effectively than NaCl (p < 0.05), Fig. 2a. Osmotherapy performed with mannitol increased the brain Na⁺ content by 6% (p < 0.001), which was less than with NaCl as the osmotic agent (15%, p < 0.001), Fig. 2b. The brain Cl⁻ content, in contrast, increased to a similar extent upon treatment with either of the osmolytes (31% with NaCl, p < 0.001 and 38% with mannitol, p < 0.001, Fig. 2c), which was also evident for the brain K⁺ content (p = 0.23; 2% with NaCl, n = 9, p < 0.01, and 3% with mannitol, n = 6, p < 0.001). Osmotherapy thus reduced the brain water content, but promoted brain electrolyte accumulation (predominantly in the form of Na⁺ and Cl⁻) irrespective of the osmotic agent employed, with NaCl being slightly more effective than mannitol for brain water extraction under our experimental conditions.

To identify the molecular mechanisms governing the hyperosmotic-induced brain ion accumulation and resulting volume regulation, the experimental regime from above (with NaCl as the osmotic agent) was repeated in rats during i.v. exposure to a mixture of inhibitors targeting a selection of ion-transporting mechanisms expressed in the BBB capillary endothelium and the blood-facing side of the choroid plexus. The diuretic compound bumetanide was applied for NKCC1 inhibition [25], amiloride to target NHE1 and ENaC [19], while the carbonic anhydrase inhibitor methazolamide [16] was applied to indirectly inhibit the NBCs. Importantly, these inhibitors did not demonstrate an effect on the arterial blood pressure of anaesthetized rats compared with vehicle (at 1 h endpoint, n = 3, Fig. 3a).

A single i.v. dose of inhibitors did not alter the plasma osmolarity compared to vehicle treatment in either control rats (vehicle: 303 ± 1 mOsm vs. inhibitors: 303 ± 2 mOsm, n = 7–9, p = 0.76) or osmotherapy-treated rats (vehicle: 355 ± 1 mOsm vs. inhibitors: 357 ± 2 mOsm, n = 8–9, p = 0.35). Delivery of inhibitors did not affect the brain water, Na⁺, and Cl⁻ content in control rats and failed to modulate the osmotherapy-induced changes in brain water, Na⁺, and Cl⁻ content, Fig. 3b–d. The K⁺ content was also unaffected by i.v. inhibitor application (in mmol/kg dry weight: control; vehicle: 463 ± 2 vs. inhibitors: 460 ± 2, osmotherapy; vehicle: 471 ± 2 vs. inhibitors: 474 ± 2, n = 7–9, p > 0.80). To increase the probability for the inhibitors to reach their targets in sufficient concentrations, we performed an additional experimental series with triple inhibitor application (20 min and 5 min prior to initiation of hyperosmotic treatment and 15 min after). These increased inhibitor doses did not affect the brain water content (Fig. 3b, inset). The unchanged electrolyte contents following inhibitor exposure aligns with the stable brain water content. These results suggest that NKCC1, NHE1, ENaC, and NBCs localized at the blood-facing side of the BBB capillary endothelium and the choroidal membrane are not the primary access routes for brain electrolyte entry during osmotherapy and therefore not the molecular mechanisms underlying brain volume regulation under these conditions.

Ion-transporting mechanisms localized at the other major interface; the ventricular side of the choroid plexus, may instead contribute to the volume regulatory gain of cerebral electrolytes upon administration of osmotherapy in the form of a hyperosmotic NaCl challenge. The select ion-transporting mechanisms expressed at the luminal membrane of the choroid plexus epithelium were targeted by injection of the inhibitor mixture (estimated ventricular concentrations of 20 μM bumetanide, 100 μM amiloride, and 100 μM methazolamide) directly into one of the lateral ventricles. Initially, the maximal inhibitor volume and infusion rate were chosen from two criteria: (1) both lateral ventricles should be exposed to inhibitors even though injections were given into only one of the lateral ventricles (verified with Evans blue, see Fig. 4a for a representative image) and (2) the ICP should remain fairly stable upon intraventricular inhibitor infusion (the ICP increased briefly to only a minor extent; 2.6 ± 0.7 mmHg, n = 3, Fig. 4b, with a brief compression of the jugular vein illustrated as a positive control). Vehicle or inhibitors were thus injected into the ventricular system of anesthetized rats prior to osmotherapy followed by another drug application every 15 min during the 1 h experimental time period to maintain a maximal targeting effect despite risk of wash-out by the high ventricular CSF flow rate [26]. The plasma osmolarity was similar in vehicle- and inhibitor-treated rats exposed to isosmolar NaCl solution (vehicle: 297 ± 2 mOsm vs. inhibitors: 298 ± 2 mOsm, n = 6, p = 0.94) and in rats subjected to osmotherapy (vehicle: 347 ± 1 mOsm vs. inhibitors: 347 ± 2 mOsm, n = 6, p = 0.87). Osmotherapy led to a reduction in the brain water content and to an increased Na⁺ (12%) and Cl⁻ (22%) content in the brain of vehicle-treated rats (n = 6, p < 0.001 for both, Fig. 4c–e), with an unaltered brain K⁺ content (in mmol/kg dry weight: control: 472 ± 3 vs. osmotherapy: 471 ± 4, n = 6, p > 0.90). Intraventricular inhibitor application had no effect on the brain water content in control rats or in osmotherapy-treated rats, Fig. 4c. The brain Na⁺ and Cl⁻ content in control- or osmotherapy-treated rats was, likewise, unaffected by inhibitor application into the lateral ventricles (n = 6, for all conditions, Fig. 4d, e), which was also seen for brain K⁺ content (in mmol/kg dry weight: control; vehicle: 472 ± 3 vs. inhibitors: 469 ± 4, osmotherapy; vehicle: 471 ± 4 vs. inhibitors: 472 ± 2, n = 6, p > 0.90 for both). These results suggest that osmotherapy-mediated brain electrolyte influx does not originate from increased activity of choroidal transporters (NKCC1, NHE1, NBCs, or ENaC) expressed at the luminal CSF-facing side of the membrane.

To assess whether Na⁺ and Cl⁻ entered the brain through a possible breach in the brain barriers in response to osmotherapy, we delivered Na⁺-fluorescein i.v. 5 min prior to osmotherapy treatment (as above). Histological analysis of coronal brain sections (Fig. 5a) from the control rats revealed a weak background fluorescent signal in the brain parenchyma, as illustrated before [23], near-absence of Na⁺-fluorescein in the neocortex, hippocampus, and thalamus, and minor staining in the lateral ventricle [23] (from choroid plexus with fenestrated blood capillaries), n = 3, Fig. 5c. Notably, the observed staining pattern was unaltered by osmotherapy treatment as illustrated in representative images of the neocortex, hippocampus, thalamus, and lateral ventricle, while no fluorescence was detected in naïve rats, which did not receive Na⁺-fluorescein (n = 3, Fig. 5b–d). The pineal gland (Fig. 5e) served as a positive control due to the lack of BBB in this brain structure. Hence, Na⁺-fluorescein was detected in the pineal gland of control rats and osmotherapy-treated rats, while no fluorescence was observed in the pineal gland of naïve rats, which did not receive Na⁺-fluorescein (n = 3, Fig. 5f–h). The absence of osmotherapy-induced penetration of Na⁺-fluorescein into the brain indicates that the integrity of the BBB and BCSFB remained intact during the applied osmotherapy treatment.