Metabolism generates a number of by-products that are not useful and could be potentially dangerous to the system. These waste metabolites exit cells into the interstitial fluid that drains into lymphatic vessels. Communication of the lymphatic system with the blood vascular system ensures that the metabolites are removed from the body by excretion. Until recently, the central nervous system (CNS) was thought to be an exception in that it managed its own waste without any communication with the peripheral circulatory systems. This is because there was no detectable lymphatic vasculature in the brain parenchyma [1]. Conversely, injection of dyes into the peripheral circulation did not appear in the CNS indicating the presence of a clear blood-brain barrier (BBB) [2], conferring an “immune privileged” status on the organ [3]. And yet, the presence of certain CNS-specific products–for example, α–synuclein–in the peripheral circulation, suggested a connection between these two systems. This review attempts to give a comprehensive account of understanding of the flow of cerebrospinal fluid (CSF) which now involves the glymphatic system [4] and the meningeal lymphatic system [5]. This has not only changed our understanding of fluid circulation in the brain but has also helped in understanding the pathology of neurodegenerative diseases, and ischemia [6]. Glymphatic failure is proposed to be a common bottleneck for proteinopathies like Alzheimer's disease, frontotemporal dementia, chronic traumatic encephalopathy, Parkinson’s disease, Lewy body disease, Huntington’s disease, and amyotrophic lateral sclerosis and a promising therapeutical target [7].

[enlarge]

Figure 1. The glymphatic system which facilitates the removal of solutes from the brain parenchyma. 1. Glial cell. 2. Astrocytic endfeet. 3. Cranial artery. 4. Cranial vein. 5. Meningeal lymphatic vessel opening (see below). 6. Subarachnoid space. 7. Brain parenchyma. 8. Lateral ventricle. 9. Third ventricle. 10. Fourth ventricle. 11. Spinal arachnoid space.

Like all other organs, the brain parenchyma secretes its waste into the interstitial fluid (ISF). This interstitial fluid makes up a fifth of the fluid in the CNS. The rest is contributed by a secretion from the choroid plexuses which are loose papillary structures formed from extensions of the ventricular ependymal tissue surrounding the cerebral capillaries and protrude into the ventricles. This cerebrospinal fluid (CSF) is mainly formed from the plasma that diffuses out of the capillaries, and is enriched with proteins synthesised in the choroid plexuses. CSF contains lymphocytes [8].

CSF bathes the entire neuropil and provides a fluid cushion to the delicate nervous tissue. Starting from the lateral ventricles, the fluid moves through the foramina of Munro to the third ventricle and thence through the aqueduct of Sylvius to the fourth ventricle. From the fourth ventricle, it can enter the foramen of Magendie to fill the spinal subarachnoid space. Part of the fluid in the fourth ventricle can flow laterally through the foramen of Luschka to enter the subarachnoid space around the cerebral cortex [9]. Reabsorption of CSF by the arachnoid granulations (formed by the invaginations of the mater into the superior sagittal sinus or SSS) due to a pressure-dependent gradient completes the circuit of the CSF [10]. In infants, the arachnoid granulations are not fully developed and the absorption of CSF by the dural venous plexuses in the dura appears to be more prominent [11]. CSF was also shown to move along the cranial nerves to the cribriform plate into the nasal sinuses [12] and along the orbital route [13].

[enlarge]

Figure 2. Sites of tracer injection into the rodent brain. 1. Lateral ventricle. 2. Third ventricle. 3. Fourth ventricle. 4. Subarachnoid space. 5. Brain parenchyma (neuropil).

In this entire sequence it appears that interstitial fluid leaves the brain parenchyma to collect in the ventricles and the subarachnoid spaces, which are reservoirs of CSF [14].

Since the last decade, however, this flow diagram of CSF has been modified to accommodate the glymphatic system and the meningeal lymphatics [15]. The former is a second pathway for CSF/ISF flow through the Virchow-Robin spaces around the penetrating arteries and veins [4]. The latter includes bona fide extracranial lymphatic vessels originating from the subarachnoid spaces and dural spaces along the perineural sheaths of the cranial nerves that leave the CNS [16].

Moreover, apart from the arachnoid granulations, CSF has also been shown to be absorbed by pia mater capillaries, retro-orbital tissues, inner ear and the spinal canal and spinal nerve roots [17] as well as the optic nerve [18].

One of the first reports of a paravascular system in the CNS emerged from the injection of horseradish peroxidase (HRP) into the lateral ventricles of the brains of cats and dogs [29]. Subsequent development of brain slices using the chromogenic substrate TMB showed that the reporter enzyme was localised in spaces around the vascular vessels. Distribution of solutes in the brain also appeared to be regulated by arteriolar pulsation [39]. This was followed by studies using radio-iodinated serum albumin (RISA) which was found to be distributed in the brain in spaces around the arterioles and venules [22].

The glia limitans sheath around the cranial arterioles and venules were described in the 1990s. This layer is part of the blood-brain barrier (BBB) formed from the astrocytic endfeet which possess tight junctions that are selective to only some solutes [40]. The sheath was studied using GFAP (Glial fibrillary acidic protein) [41-43], S-100b [41] and aquaporin AQ4 [43] as marker molecules for confocal microscopy. Electron microscopy revealed that the sheath formed encloses a fluid filled space between the endothelial wall and the glial cell surface. This paravascular space was also shown to be continuous with the subarachnoid space [44].

Mechanisms regulating CSF transport via intracranial cavities are not entirely understood. With regard to glymphatic-lymphatic coupling and mechanisms of CSF antigen clearance, high-resolution microscopy has revealed that arachnoid granulations, known to transport CSF into venous sinuses, are formed by multiple internal channels connected to perisinus spaces [45], suggesting that arachnoid granulations contain outer capsule and inner core area and act as transarachnoidal flow passages.

In 2012, Iliff et al published their seminal paper on a paravascular pathway that allows CSF to reach into the brain parenchyma. Since the space around the penetrating cranial arteries and veins is formed by the interdigitations of the glial cells around the vasculature, the pathway was named the ‘glymphatic system’.

This pathway consists of 3 elements (Fig 1):

1. a para-arterial CSF influx route,
2. a para-venous ISF clearance route,
3. a transparenchymal pathway via the astrocytic aquaporin-4 (AQP4) water channel.

High-resolution 3D magnetic resonance (MR) non-contrast cisternography, diffusion-weighted MR imaging (MR-DWI), intravoxel-incoherent motion (IVIM) DWI and dynamic contrast-enhanced MR imaging have been used to investigate the impairment of glymphatic function induced by the deletion of AQP4 [46]. AQP4 KO mice showed larger interstitial spaces and brain volumes and reduced CSF space volumes compared to wildtype mice. The larger interstitial fluid volume correlated with reduced glymphatic influx. The impaired brain fluid transport in AQP4 KO mice may be induced by a reduction in glymphatic clearance followed by stagnation of fluid in the interstitial space.

Introduction of a dye, an enzyme, a fluorescent or radioactive tracer into the brain has been the method of choice to study the flow of solutes in the CNS [47].

Tracers can be injected into the brain at various sites (Table 2). The advantage of injection into the cisterna magna (intracisternally) is that cannulation can reach down to the mid-brain without damaging the neuropil. It also allows injection of small aliquots of the tracer over a period of time to study the flow of CSF [26]. In contrast, injection of tracers into the ventricles involves the penetration of the brain parenchyma and subsequent damage. Even cortical injections can damage the delicate neuropil. Likewise, injection into the subarachnoid spaces results in changes in intracranial pressure and altered flow of CSF (Fig 2).

Plasma concentrations of neurodegeneration biomarkers were found to associate with the parameters measuring glymphatic and meningeal lymphatic systems [52]. The study evaluated the potential corelation between plasma biomarkers, 40 and 42 amino acid-long amyloid-β (Aβ40 and Aβ42), total-tau, glial fibrillary acidic protein (GFAP), and neurofilament light (NfL), and magnetic resonance imaging measures of CSF-mediated clearance from brain and CSF-to-blood clearance parameters of pharmacokinetic modeling. Overall, there was individual- and disease-specific association of neurodegeneration biomarkers’ concentration in plasma with the functional parameters of glymphatic and meningeal lymphatic systems.

Instead of single injections, convection enhanced delivery (CED) which employs an infusion pump is a better method to study the distribution of solutes in the brain parenchyma [19]. This is because infusion at a steady pressure can ensure a greater distribution of the reporter [4].

Evans blue is the classic albumin binding dye that is used to study the flow in the vascular system [55]. Similarly, dextran is a preferred tracer for the study of lymphatic vessels [28]. With the availability of a large choice of fluorescent labelled markers, it is now convenient to label and study the movement of proteins (albumin) and large carbohydrates (dextran, inulin) within the CNS. Table 2 lists the tracers that have been used to study the flow of CSF in the brain.

Earlier workers used light microscopy or electron microscopy to ascertain the distribution of dye-coupled dextrans and enzyme reporter systems [27, 29] or fluorescence imaging (for fluorescent labelled molecules) of brain slices.

Two photon laser scanning microscopy has also been used to trace the path of fluorescent labelled dextran in real time through a cranial window opened above the brains of anesthetised mice [4]. Multiphoton fluorescence involves the simultaneous absorption of two or more multiple long wavelength photons by the fluorophore to reach the excited state. The absorption has an additive effect giving a magnitude equivalent to the sum of the individual photon energies. This allows a better penetration, less photo- damage and no out-of-focus bleaching without altering the emission spectrum [65].

For radioactive tracers, the brain is harvested without the dura at various time points after the injection of the tracer and solubilised and total brain radioactivity measured by liquid scintillation. The activity does not include the tracer in the subarachnoid compartment.

Transgenic mice where endothelial cells expressed fluorescent labelled reporter molecules can be used to enhance the clarity of tracer distribution. For e.g. FVB/N-Tg(GFAPGFP)14Mes/J (GFAP-GFP, JAX) mice which express GFP under the astrocytic endfeet marker protein GFAP offers a good contrast to a red fluorescent tracer. Also, the role of aquaporin 4 in the glymphatic system was confirmed using a mouse KO without AQ4 expression [4].

Magnetic resonance imaging has been successfully used to validate the presence and working of the glymphatic system in rats [66] and humans [67]. A specialised method of MRI called diffusion tensor imaging (DTI) has also been successful in tracing the functioning of the glymphatic system [68]. This technique is based on the measure of anisotropy of water that acquires a directionality of flow when channelized into specific paths and does not require the injection of a contrast dye.

Within minutes of the injection of a tracer molecule into the cisterna magna of different animals, it can be detected in the lymph nodes. Likewise, a tracer injected into the brain parenchyma localises in the basement membranes associated with the endothelial cells or the pial-glial membrane [69].

The permeability of the system is limited to small molecules as seen by the confinement of a large molecular weight tracer ( e.g. dextran 2000) to a paravascular path while low molecular weight A594 and Tr-dextran 3 moved into the brain parenchyma and were distributed uniformly [4].

This was further confirmed by the rapid entry of ³H–mannitol into the parenchyma and the slow movement of ³H–dextran in the paravascular space [4].

Injections into the cortex showed that the tracers moved along the outer margins of the arterioles and venules. Again, a small molecule like TR3 was found to cross into the interstitium while the larger FITC-dextran2000 remained in the paravascular space that is continuous with the subarachnoid space [70].

With regard to the flow of CSF through the paravascular (glymphatic) pathway, movement of adenoviral capsids and fluorescent labelled liposomes suggested that CSF does enter the brain parenchyma. The forward movement of these reporter systems was shown to depend on the pulsatile motion of the arterial wall [19]. Studies have now revealed that breathing patterns [67], exercise, the position of the head [56], sleep [71, 72] and even myelination [73] play a role in the correct functioning of the system.

Notably, influencing the noradrenergic control of central glymphatic flow may be effective for treating acute traumatic brain injury [74]. Impaired glymphatic and lymphatic fluid flow triggered by excessive norepinephrine release (adrenaline storm) is one of the known causes of acute brain edema following injury. Using a mouse model of traumatic brain injury, the authors have shown that pan-adrenergic receptor inhibition corrected central venous pressure and partly restored glymphatic and cervical lymphatic flow reducing brain oedema and normalizing functional outcomes. Furthermore, inhibition of adrenergic signaling after brain injury has also normalized fluid flow and stimulated debris clearance followed by reduced inflammation.

In addition, NOTCH3 was found to be involved in the regulation of glymphatic flow [75]. In particular, aging-induced decline in Notch3 signaling in both murine and human brain vessels was associated with impaired vascular reactivity including dilation, tortuosity, microaneurysms and decreased cerebral blood flow. Combined, these vascular changes negatively affected glymphatic flow and caused accumulation of glycosaminoglycans within the brain parenchyma. Furthermore, single-cell RNA sequencing of the neuronal compartment in aging Notch3-null mice showed features similar to those found in patients with neurodegenerative diseases.

Thus, the glymphatic system denotes a network of perivascular channels through which an exchange of CSF and ISF takes place in the brain interstitium [4]. This not only facilitates the clearance of interstitial solutes [76] but also allows the entry of CSF into the dense parenchyma.

As mentioned earlier, the absence of detectable lymphatic vessels led to the belief that the CNS managed its own waste. And yet, a successful immune response was mounted against sheep erythrocytes introduced into the ventricular space of rabbits [77]. Experimental evidence of interstitial fluid draining into the deep cervical lymph nodes also suggested the involvement of extracranial lymphatics [21, 78].

Experimental evidence for the existence of the meningeal lymphatic system came from scanning electron microscopy of the meninges [79]. Round and oval “stomata” were found to be scattered in the mesothelial cells of the meninges which possibly connected to the cervical lymphatic system. The reason for the detection of openings and not whole lymphatic vessels was because the conventional method of studying the brain did not include the dura mater [35]. In fact, the dura is so closely associated with the cranium, that it is advisable to solubilise the bone in order to recover the brain along with all three meninges intact. Louveau et al (2018) [16] described the presence of a lymphatic system that emerges from the dural sinuses and continues into the deep cervical lymph nodes.

The method of choice to study the lymphatic system is to inject a tracer and detect its presence in brain slices after a few minutes. Fluorescent tracers are preferred since dyes such as Evan’s Blue can dissociate from the coupled protein and cannot be detected in histological preparations. Microfil injections–which are performed post mortem and under pressure–can damage the fragile meningeal lymphatics [49].

In order to establish the lymphatic nature of the vasculature, immunofluorescence has been used to detect the classical lymphatic vessel markers such as LYVE-1and podoplanin. Double immunofluorescence with anti-CD31and Prox1 has also been used [49]. The meningeal lymphatic system has been extensively studied in human cadavers also [50]. In mice, meningeal lymphatic vessels can be ablated through the photodynamic dye Visudyne [47].

[enlarge]

Figure 3. The current understanding of CSF flow in the central nervous system.

Typically, lymphatic vessels are characterised by the presence of fenestrated endothelium, and absence of a basal membrane and luminal blood cells. The meningeal lymphatics do resemble the peripheral lymphatics except that there are no valves in these vessels to prevent a back flow (Table 4).

CSF was shown to leave the subarachnoid space of the CNS through lymphatics along the cranial retroglenoid vein and sigmoid sinus as well as the dural side of the pterygopalatine artery [5]. This was further confirmed by using K14-VEGFR3-Ig mice, which have complete aplasia of lymphatic vasculature [80]. The outflow of CST to deep cervical lymph nodes in mice can also be measured [47].

The meningeal lymphatic system, however, does not share any features with the glymphatic system. Some of the differences between the two are listed in Table 5.

With the discoveries of the glymphatic system and the meningeal lymphatic systems, it became imperative to re-trace the path of CSF circulation in the CNS [1, 81]. Another reason to revisit CSF circulation is because the original assumption of CSF absorption by the arachnoid granulations does not apply to neonates. In humans, while CSF secretion from the choroid plexus starts by the third month of gestation, arachnoid villi are not developed until after birth [82].

It has now been suggested that the CSF enters the brain interstitium via the subarachnoid space. A mix of ISF and CSF drains out through the paravenous route. It then enters the meningeal lymphatics and is drained into the deep cervical lymph nodes (dCLNs). This, in turn, can connect to the inner carotid and empty the waste metabolites into the general circulation.

A recent study has shown that the removal of metabolic waste and the supply of metabolites were enhanced by functional hyperemia [83]. The whisker stimulation upregulated both glymphatic influx and clearance in the mouse somatosensory cortex with a 1.6-fold increase in periarterial CSF influx velocity. Particle tracking velocimetry demonstrated a coupling between vascular dilation /constriction and periarterial CSF flow velocity. Furthermore, optogenetic manipulation of vascular smooth muscle cells stimulated glymphatic influx without neural activation.