Aquaporin-4 and brain edema

In the normal adult brain, water is distributed between several compartments, including CSF (CSF, ~75–100 ml), blood (~75–100 ml) and intracellular (1,100–1,300 ml) and interstitial (~100–150 ml) brain parenchyma. Water moves between the different compartments in response to osmotic gradients and hydrostatic pressure differences. Because the brain is enclosed in a rigid skull, brain swelling produces displacement of water from low-pressure compartments, including CSF and venous blood (~10 mmHg), into peripheral blood. Once the low-pressure reserve is exhausted, intracranial pressure progressively rises; the sulci become effaced on computed tomography (CT) and magnetic resonance image (MRI) scans, then the third ventricle becomes invisible, and finally the basal cisterns disappear due to brain-stem herniation. High intracranial pressure can therefore cause brain ischemia, herniation, and death. This scenario, however, does not apply to young children (less than a year old) in whom the open fontanelles can accommodate, to some extent, the swollen brain. In this situation, the fontanelles bulge and the head circumference increases, minimizing the rise in intracranial pressure. However, even the open fontanelles cannot compensate for rapid brain swelling, and therefore, young children are not fully resistant to the development of high intracranial pressure and brain herniation.

Brain edema management includes sedation and avoidance of hypercapnia to prevent intracranial pressure elevation, administration of intravenous hyperosmolar solutions such as mannitol and hypertonic saline, corticosteroids for brain tumors, surgical resection of the causative lesion, and in extreme cases, decompressive craniectomy. For critically ill patients, invasive monitoring of intracranial pressure and cerebral perfusion pressure is done to optimize therapy. However, many of these therapies to reduce brain swelling were introduced early in the early/mid twentieth century, and their efficacy is limited [17, 18]. The paucity of effective drugs to be used in brain edema reflects, in part, the incomplete understanding of cellular mechanisms involved in brain edema formation and resolution.

A logical framework for understanding the cellular mechanisms of brain edema was first suggested by Igor Klatzo in the 1960s, when he classified edema into cytotoxic and vasogenic types; a third type, termed hydrocephalic edema, has subsequently been described [15]. Common causes of brain edema are summarized in the Table 1.

Cytotoxic edema, as seen in hyponatremia and early cerebral ischemia, is intracellular accumulation of water due to energy failure and inability of cells to regulate their volume [16]. This results in a shift of water from the interstitial into the intracellular compartment and a net uptake of water from the blood compartment into the brain parenchyma. Astrocytes are the main cell type that swell in cytotoxic brain edema, especially the pericapillary foot processes [16], which are the predominant sites of AQP4 expression in the brain [8, 19]. Astrocyte swelling may be an important early event predisposing the brain to further damage. Because cytotoxic edema affects all cells, both gray and white matter swell, resulting in loss of the clear margin between gray and white matter on CT and MRI scans. Because the blood–brain barrier is intact, excess brain water is not accompanied by protein, and there is no brain enhancement on CT or MRI scans after intravenous contrast administration.

Vasogenic edema, of which brain tumor and brain abscess edema are prime examples, involves disruption of the blood–brain barrier. As a result, iso-osmotic fluid and serum proteins enter the interstitial space from the bloodstream in response to a hydrostatic pressure difference. Therefore, in vasogenic brain edema, there is expansion of the interstitial space [20]. The resistance to flow of interstitial fluid is higher in gray matter (consisting of tangles of cell processes) compared with white matter (primarily consisting of aligned neuronal tracts), which explains why vasogenic edema fluid is found in the white matter on CT and MRI. Because the blood–brain barrier is open, vasogenic brain edema is associated with enhancement of the causative lesion (such as brain tumor) on CT or MRI scans after intravenous contrast administration.

Despite the success of the Klatzo classification system, most clinical conditions consist of mixtures of different types of edema occurring with different time courses. For example, early cerebral ischemia produces cell swelling, but later on, when the capillary endothelium becomes damaged, the blood–brain barrier is disrupted, resulting in vasogenic-type swelling.

The mechanisms of edema fluid clearance are less well understood than the mechanisms of edema fluid formation. In all brain edema types, excess fluid leaves the brain parenchyma along three different routes: across the blood–brain barrier into the bloodstream, across the ependyma into the ventricles, and across the glia-limiting membrane into the CSF in the subarachnoid space (Fig. 2). CSF eventually enters the arachnoid granulations and is cleared into the superior sagittal venous sinus.

The relative contributions of the three major routes of edema fluid elimination from the brain are not known. It has been suggested that vasogenic edema fluid is cleared primarily by bulk flow through the extracellular space and across the glia limitans into the CSF [21]. This idea is based on old experiments that measured clearance rates of inert dyes injected into brain extracellular space. The dyes were primarily eliminated into the CSF at equal rates independent of molecular weight, favoring a bulk-flow mechanism [21]. A major flaw with this interpretation is the unjustified assumption that dye and water efflux routes are identical. More recent data suggest that excess brain water elimination in vasogenic edema across the glia limitans involves a transcellular route, not bulk flow [22].

Death of neurons and glia in cytotoxic edema releases their intracellular contents into the extracellular space. Regulatory volume decrease is an interesting phenomenon whereby swollen cells release intracellular ions (mainly K⁺ and Cl⁻) and amino acids (such as taurine and glutamate) into the extracellular space [16]. This produces a decrease in cell size toward baseline. It is possible, then, that in cytotoxic edema, the excess water initially resides in the intracellular compartment but ultimately moves to the extracellular space due to cell death and regulatory volume decrease. Therefore, excess water elimination routes in cytotoxic edema may be the same as those in vasogenic edema.

The first evidence that AQP4 plays a major role in cytotoxic brain edema came from studies with AQP4-null mice [23]. Intraperitoneal water injection causes profound hyponatremia to 105 mM within 5 min and subsequent death from brain swelling and increased intracranial pressure. Mortality after hyponatremia is markedly reduced in AQP4-null mice (Fig. 3). This protection is associated with reduced blood–brain barrier water permeability and reduced rate of water flow into the brain parenchyma. AQP4-null mice are also protected from other models of cytotoxic brain edema, including bacterial meningitis [23] and early focal cerebral ischemia [24]. AQP4 protein in plasma membranes is thought to be bound on an aggregate of intracellular proteins, including α-syntrophin [25]. Interestingly, α-syntrophin-null mice are also protected from cytotoxic brain edema produced by focal cerebral ischemia [26]. Taken together, these findings suggest that drugs that inhibit AQP4 expression or function might limit cytotoxic brain swelling in humans.

An unexpected role for AQP4 in vasogenic brain edema was shown using three models of vasogenic brain edema in mice [22]. AQP4 null mice had more brain swelling compared with wild-type mice after cortical freeze injury and brain tumor implantation. The observation that intraparenchymal infusion is eliminated more slowly in AQP4-null mice compared with wild-type mice suggests that excess brain water elimination in vasogenic edema is defective after AQP4 deletion. Once the excess interstitial fluid (vasogenic edema) reaches the barriers between brain parenchyma and fluid compartments, it must be eliminated by a transcellular AQP4-dependent route. These findings suggest that AQP4 activators or upregulators, when available, may reduce vasogenic brain edema in humans.

Hydrocephalus is the accumulation of CSF with ventricular enlargement due to obstruction in CSF flow (obstructive or noncommunicating hydrocephalus) or impaired CSF absorption (communicating hydrocephalus). Causes of obstructive hydrocephalus can be congenital, e.g., aqueduct stenosis preventing CSF flow from the third to the fourth ventricle; or acquired, such as brainstem glioma or intraventricular hematoma. Communicating hydrocephalus is often associated with damage to the arachnoid granulations, as in meningitis. In a model of obstructive hydrocephalus produced by injecting kaolin in the cisterna magna of mice, AQP4-null mice had accelerated ventricular enlargement progression compared with wild-type mice (Fig. 4) [27]. Presumably, the reduced water permeability of the ependymal layer, subependymal astrocytes, and glia limitans produced by AQP4 deletion reduces the elimination rate of CSF across these borders into the subarachnoid space.

Apart from its role in brain edema, AQP4 was found to a play major role in cell migration [28‐31] and neural excitability. Astrocyte migration is delayed in AQP4-null mice, and as a result, glial scarring is impaired [29, 31]. AQP4 is thought to accelerate astrocyte migration by facilitating transmembrane water flow, which accompanies the fast changes in cell shape that occur during migration. Several lines of evidence show reduced neural excitability in AQP4 deletion: AQP4-null mice have a higher seizure threshold and prolonged seizure duration than do wild-type mice [32, 33], and they have sensorineural deafness [34] and mild retinal impairment [35]. The mechanism by which AQP4 influences neural excitability is unknown but may involve interaction between AQP4 and an astrocytic potassium channel (Kir4.1) [36], or altered extracellular space size [37].