Understanding the development, pathogenesis, and injury response of meningeal lymphatic networks through the use of animal models

Abstract

Meningeal lymphatic vessels (MLVs) help maintain central nervous system (CNS) homeostasis via their ability to facilitate macromolecule waste clearance and neuroimmune trafficking. Although these vessels were overlooked for centuries, they have now been characterized in humans, non-human primates, and rodents. Recent studies in mice have explored the stereotyped growth and expansion of MLVs in dura mater, the various transcriptional, signaling, and environmental factors regulating their development and long-term maintenance, and the pathological changes these vessels undergo in injury, disease, or with aging. Key insights gained from these studies have also been leveraged to develop therapeutic approaches that help augment or restore MLV functions to improve brain health and cognition. Here, we review fundamental processes that control the development of peripheral lymphatic networks and how these might apply to the growth and expansion of MLVs in their unique meningeal environment. We also emphasize key findings in injury and disease models that may reveal additional insights into the plasticity of these vessels throughout the lifespan. Finally, we highlight unanswered questions and future areas of study that can further reveal the exciting therapeutic potential of meningeal lymphatics.

Introduction

The recent (re)discovery of meningeal lymphatic vessels (MLVs) has uncovered a novel vascular network integral to central nervous system (CNS) homeostasis. The resulting paradigm shifts regarding our classical understanding of CNS fluid circulation, macromolecule drainage, and neuroimmune regulation brought about by this system have caused these vessels to become heavily investigated and scrutinized. Utilizing rodents, pathogenic changes to MLVs have now been identified in several models for neurological diseases, craniofacial disorders, and aging, igniting further curiosity about how these vessels develop, function, and integrate with known systems in the head.

MLVs are unique in comparison to peripheral lymphatic vessels; in addition to their structural properties and postnatal course of development beneath the skull [1], over 300 genes are differentially expressed in MLVs [2]. These findings underscore the need to identify factors that influence their development within the unique meningeal environment, and how these vessels synergize with the development of other systems (i.e., the glymphatic system). However, currently existing animal models used to study MLVs are often focused on adult-onset disease or injury, and there are few models which specifically shed light on the onset and regulation of meningeal lymphangiogenesis.

Here, we review current understandings stemming from classical lymphatic development in the periphery and their potential application to the developmental trajectory of MLVs. We discuss important signaling mechanisms, transcription factors, and events in the developmental time-course of MLVs, and how these overlap with skull growth and onset of cerebrospinal fluid (CSF) circulation in the head. We also highlight animal models permitting investigation of meningeal lymphatic function, regulation, and alteration in disease, with a focus on how the unique meningeal environment may shape MLV development and maintenance. Finally, we provide a critical review of questions, challenges, and directions for future study.

Mechanisms of lymphatic development derived from the periphery

Much of our understanding of mechanisms that control the growth, remodeling, and maintenance of lymphatic vessels is primarily derived from work on peripheral lymphatics. Collectively, these studies have shown that lymphatics are predominantly venous-derived vessels that originate from the anterior cardinal vein (ACV) during mid-embryogenesis [3]. Starting at embryonic (E) day 9.5 (E9.5) in mice, a subpopulation of endothelial cells in the ACV start to express Prox1, the master transcription factor of lymphatic identity which specifies these endothelial cells to adopt a lymphatic endothelial cell (LEC) fate [4, 5]. The specific cues from the venous environment necessary for the induction of Prox1 expression are not entirely understood [5]. However, COUP-TFII, a transcription factor expressed in veins but not arteries, is necessary for Prox1 expression, supporting the generally accepted idea that a significant portion of peripheral lymphatics are of venous origin [6, 7].

Apart from venous-derived lymphatic vessels, an ever-expanding number of animal studies has uncovered peripheral lymphatic networks with non-venous origin. For example, experiments in mice with Prox1 inactivation using the Tie2-Cre driver, deleting Prox1 in venous-derived LECs, suggest that some dermal lymphatic vessels in the dorsal and lumbar midline form through the assembly of a subpopulation of non-venous, non-Tie2 lineage cells [8]. Additionally, studies have identified a potential role of Isl1-expressing pharyngeal mesoderm in forming cardiac LECs in mice [9]. These findings are not limited to rodents; in zebrafish, a population of facial lymphatic progenitors is found to sprout from a migratory cell lacking both venous and lymphatic identity cell markers [10]. Although not well understood, the discovery of what is now termed “lymphvasculogenesis,” or the differentiation and coalescence of lymphatic networks arising from non-venous single cell precursors, continues to change our current understanding of peripheral lymphatic development [11].

Both venous and non-venous-derived immature LECs express genes that are characteristic of lymphatic identity, such as Lyve1, Nrp2, and Vegfr3 [4, 12, 13]. The expression of Vegfr3 on the surface of immature LECs renders these cells responsive to lymphangiogenic mitogens, such as Vegf-c and Vegf-d [4]. Between these two cognate ligands, Vegf-c is the major factor implicated in lymphangiogenesis [14–16]. Ligand–receptor interactions between Vegf-c and Vegfr3 are necessary for lymphatic-specified cells to bud off the ACV, migrate, and form primitive lymphatic structures known as primary lymphatic sacs [17]. These sacs are visible at E12.5 and act as a reservoir of LECs throughout the developing embryo [4]. Immature LECs bud and sprout from primary lymphatic sacs until E14.5, by which time they have finished spreading throughout the entirety of the embryo [5]. These LECs then differentiate and remodel by the end of embryogenesis to form mature lymphatic networks with the capability of draining fluid.

Within peripheral tissues, functional lymphatic networks are organized under a stereotyped hierarchy. Interstitial fluid is first absorbed by lymphatic capillaries. These are oak-shaped, blind-ended vessels which comprise the most distal portion of the lymphatic network and are positive for Prox1, Lyve1, PdpN, and Vegfr3 [18–20]. These vessels uptake interstitial fluid, soluble antigens, macromolecules, and lipids [20]. Lymphatic capillaries also have specialized features to mediate fluid absorption and immune cell uptake, such as “button-like junctions” composed of adherens and tight junctions which are dependent on Vegfc–Vegfr3 interactions for their development [21–23]. Due to the discontinuous nature of these junctions, gaps between LECs of lymphatic capillaries enable the movement of antigen-presenting cells (APCs) into the lymphatic capillary lumen [24]. Additionally, these vessels express chemokines (e.g., CCL21) to enable APC homing to the lymphatic endothelium [25, 26].

Downstream of lymphatic capillaries, fluid flows into vessels called pre-collecting lymphatics or ‘precollectors’ which contain one-way valves [27]. These intermediate vessels share characteristics of both lymphatic capillaries and lymphatic collectors, terminal vessels of the lymphatic network that bring fluid back into venous circulation [28, 29]. For example, collecting lymphatics are uniquely characterized as having a smooth muscle coating to help propel fluid movement. They are also connected by continuous zipper-like junctions that, unlike button junctions, provide a tight seal [30]. However, pre-collecting lymphatics have both zipper-like and button-like junctions and lack smooth muscle cells [29]. These observations highlight how different subsets of LECs have unique morphological features that are related to their identity and specialized for different aspects of immune cell uptake and waste clearance.

Development and maintenance of meningeal lymphatics

Unlike most peripheral lymphatic vessels, lymphatic vessels within the dura mater of the meninges develop postnatally. This feature is not necessarily unique to meningeal lymphatics, as lymphatics in the small intestine, ear skin, and pericardium also develop after birth [31, 32]. However, given that most lymphatics form and remodel embryonically, postnatal growth of MLVs within their specialized meningeal environment is suggestive of unique aspects to their development.

Meningeal lymphatic networks develop in a stereotyped manner in mice (Fig. 1). Lymphatic vessels first enter the cranium through skull foramina, with growth commencing from the skull base and gradually progressing to the dorsum of the head throughout the first postnatal month [29, 33]. Meningeal lymphatics grow along the dural venous sinuses and are present along the sigmoid sinus by postnatal (P) day 8 (P8) [1] (Fig. 1a). These vessels continue growing dorsally along the transverse sinuses, reaching the confluence of the sinuses by P16, with growth then proceeding rostrally adjacent to the superior sagittal sinus (SSS) [1]. By P20, dural venous lymphatics are approximately halfway along the SSS, and by ~ P28, these lymphatics converge with a separate subset of dural lymphatic vessels originating from the area around the rostral rhinal vein to form a continuous network [1] (Fig. 1a). Concurrent with lymphatic growth along the dural venous sinuses, a subset of lymphatics grows postnatally along the dural arteries. At P4, an initial network of periarterial lymphatics is present along the entirety of the pterygopalatine artery (PPA). As postnatal development progresses, this network progresses dorsally along the middle meningeal artery (MMA) toward the top of the cranium [1]. Throughout this period, periarterial lymphatics remodel and some LECs near the intersection of the PPA and the MMA form valves. By ~ P28, the periarterial network has finished developing and growth along the entire length of the MMA is complete [1].

Fig. 1.

MLVs grow along dural venous sinuses and require Vegfc–Vegfr3 signaling for their development and maintenance. A LECs enter the head through foramina at the base of the skull, organize into meningeal lymphatic vessels, and grow along dural venous sinuses. MLVs reach the jugular veins by P3, the transverse sinuses by P8, the superior sagittal sinus by P20, and the rostral-most portion of the superior sagittal sinus by P28. A′ Vegfc is expressed in venous smooth muscle, which binds to Vegfr3 receptors on MLVs, facilitating lymphangiogenesis. B In a Vegfc^flox/flox conditional knock-out, Vegfc levels are reduced by 50% and MLVs are hypoplastic, indicating that Vegfc is required for developmental lymphangiogenesis. C When Vegfr3 signaling is inactivated during adulthood, animals display regression of MLVs within 1–2 weeks, suggesting that Vegfc–Vegfr3 signaling is also required for long-term maintenance of meningeal lymphatic networks

Similar to peripheral lymphatics, the expansion of meningeal lymphatic networks is dependent upon paracrine Vegf-c signaling, which stimulates lymphangiogenesis by activating Vegfr3. Mice with conditional Vegf-c deletion by a Rosa26-CreER^T2 driver during early or midstages of meningeal lymphangiogenesis lack MLVs [1] (Fig. 1b, inset). Furthermore, mice heterozygous for a Vegf-c null allele also lack meningeal lymphatics, while Vegf-d^null mice have intact networks, suggesting that the development of these vessels relies principally on Vegf-c [1]. In adult mice, inactivation of Vegfr3 signaling with either a Vegfc/d trap or Vegfr3 blocking antibodies results in hypoplastic meningeal lymphatic networks [1] (Fig. 1c, inset). It has also been shown that dorsal vessels along the transverse and superior sagittal sinuses may be more sensitive to the levels of Vegf-c compared with basal vessels along the sigmoid and petrosquamosal sinuses [29]. Taken together, these data suggest that Vegfc–Vegfr3 signaling is indispensable for both meningeal lymphatic development and maintenance, but the requirements for Vegfc–Vegfr3 signaling may differ between dorsal and basal MLVs.

There are several potential sources of Vegf-c in the meninges and surrounding tissues. Whole mount dural preparations from Vegf-c^LacZ mice have revealed LacZ expression in smooth muscle coating the dural venous sinuses and meningeal arteries (Fig. 1a’, inset), and also the pineal and pituitary glands [1]. Meningeal lymphatic growth along the venous sinuses and meningeal arteries coincides with the development of venous/arterial smooth muscle and onset of Vegf-c expression, suggesting that smooth muscle could be the main source of Vegf-c necessary for meningeal lymphangiogenesis. Dural macrophages could also be involved in the growth and expansion of MLVs, although this needs to be tested in animal models. Interestingly, the presence of Lyve-1/CD206(+) dural macrophages precedes the growth of lymphatics along the arteries and venous sinuses, and these cells greatly decline in number once MLVs grow and expand into mature networks [1]. Macrophages can secrete Vegf-c and have been shown to mediate lymphangiogenesis in a variety of peripheral organs (e.g., the heart or the bladder) [34, 35]. Thus, it is possible that dural macrophages are involved in meningeal lymphangiogenesis, and further studies are needed to determine their exact role(s), if any, and whether these cells could be a potential source of growth factors.

Beyond Vegf-c/Vegfr3 signaling, little else is known about the cellular and molecular mechanisms that drive meningeal lymphangiogenesis. EdU-birthdating experiments have revealed a significant number of newborn LECs in expanding networks [1], suggesting that the formation of these vessels relies on local proliferation from pre-existing networks. Peripheral lymphatics can also expand through the migration of LECs [36], and LEC migration can occur through both Vegf-c dependent and independent mechanisms [37, 38]. Thus, it is possible that meningeal lymphatics could form through a combination of both proliferation and migration. However, it is not clear whether LEC migration also plays an extensive role in meningeal lymphangiogenesis beyond the early migration of LECs through skull foramina [29]. Additionally, MLVs show numerous sprouts between P16 and P25 [1, 39], suggesting signaling pathways known to regulate sprouting lymphangiogenesis in peripheral tissues (e.g., Piezo1 signaling) may also be necessary for growth and remodeling in the meninges.

Although most peripheral lymphatic vessels are derived from venous endothelium, the cellular ontogeny of MLVs is still unclear. Given that MLVs enter the meninges through the skull foramen [1, 29], these vessels might be derived from either a common or separate venous source(s) outside of the skull. At P16, however, discrete clusters of LECs were observed to extend filopodia into developing lymphatic networks along the transverse sinus [1]. This suggests that at least a portion of meningeal LECs may be derived de novo from a source within the meninges (or cranium). Furthermore, similar to primordial LECs derived from the anterior cardinal vein, it is possible that a subset of LECs are specified from dural venous sinus endothelial cells. Indeed, during the first postnatal week, some blood vessels express Prox1 in regions where MLVs form [1, 33]. It is also possible that some LECs could be derived from a non-endothelial origin. Lineage tracing with Pax3-Cre:Rosa26^tdTomato and Myf5-Cre:Rosa26^tdTomato mice revealed that some MLVs are possibly derived from paraxial mesoderm [40]. This suggests that at least a subset of meningeal LECs in adult mice may be derived from mesenchymal cells and/or have a myogenic lineage, as Myf5-Cre is not known to be expressed in endothelial cells.

Importantly, meningeal lymphatics can exhibit some anatomical differences depending on their location within the skull. Collectively, lymphatic vessels growing alongside the transverse and superior sagittal sinuses are referred to as dorsal MLVs, whereas those that grow along the sigmoid and petrosquamosal sinuses are referred to as basal MLVs. Basal lymphatics in mice have been shown to consist of two distinct populations of LECs. The distal-most portion of basal lymphatics has lymphatic capillary-like characteristics, consisting of button-like junctions [29]. These capillary-like vessels are contiguous with a second, more downstream population of vessels that have intraluminal valves and a pre-collector-like phenotype [29]. By contrast, dorsal lymphatics at first glance appear to take on a purely capillary-like phenotype due to their lack of valves. One of the first descriptions of meningeal lymphatics determined that, like capillary lymphatics, dorsal lymphatics have predominantly button-like junctions [41]. However, findings from a later study reported that mature dorsal lymphatics have predominantly zipper-like junctions [29]. It is possible that dorsal lymphatics express a combination of button and zipper-like junctions and thus have a unique identity with both capillary and pre-collector characteristics. These morphological distinctions also suggest that fluid and immune cell uptake differs between dorsal and basal networks [29]. It is unclear how these vessels assume either capillary-like or pre-collector properties, as the underlying mechanisms that drive the differentiation of these subpopulations are unknown. Understanding how these morphological differences arise is important as it can help differentiate normal variation from pathological changes to network development.

Meningeal lymphatics in injury and disease models

Given that MLVs are now recognized to be essential for CNS homeostasis, their involvement in head injury and neurological disease has been extensively explored in recent years. Several studies utilizing animal models have sought to answer questions pertaining to how MLVs react to pathological changes in their environment. It is known that MLVs undergo age-related morphological changes and impairments in fluid drainage [29, 42]. Thus, investigating how deficits in meningeal lymphatic drainage exacerbate CNS pathology may further our understanding of age-associated risk factors for neurological disease.

Animal models have revealed important insights into pathological changes that affect MLVs following stroke, traumatic brain injury (TBI), and viral infection. For example, when cortical infarction is induced via photo-activation of a light-sensitive dye in a photothrombosis model of stroke, MLVs sprout atypically from the sagittal sinus into an alymphatic zone over the right sensorimotor cortex [43]. This form of ectopic lymphangiogenesis is attenuated in Vegfr3^wt/null animals which also underwent photothrombosis, suggesting that Vegfr3 signaling is necessary for the lymphangiogenic response to stroke [43]. Interestingly, this type of abnormal growth is not seen when stroke is induced by transient middle cerebral artery occlusion (tMCAo) [43]. Notably, the photothrombosis model of stroke leads to meningeal ischemia, while the tMCAo model does not [43]. Hence, it is possible that MLVs are selectively responsive to ischemic conditions in the meningeal environment, responding with compensatory lymphangiogenesis in a Vegfr3-dependent manner.

Similar findings regarding the response of MLVs to abnormal surrounding states have been seen in a mild, closed-skull mouse model of TBI. Two hours following TBI, mice injected with fluorescent beads into the cisterna magna exhibited reduced perfusion to specialized lymphatic “hotspot” regions along the transverse sinus, as well as significantly less beads in the deep cervical lymph nodes [44]. Interestingly, these mice showed increased coverage of dorsal lymphatic networks 1-week post-TBI, with a greater number of loops and sprouts, indicative of injury-induced lymphangiogenesis [44] (Fig. 2a, Table 1). Coincident with this finding, differences in CSF uptake between sham and TBI mice were no longer apparent 1-week post-TBI, suggesting that compensatory lymphangiogenesis may have helped restore functional drainage [44] (Table 1). MLVs also experience atypical expansion in a model of CNS viral infection; however, rather than restoring MLV function, heightened lymphangiogenesis in this model is associated with impaired drainage of macromolecules to the dCLNs [45], indicating that the functional implications of meningeal lymphatic expansion are variable and nuanced according to the pathological context (Table 1). Taken together, these findings from stroke, TBI, and viral infection mouse models reinforce the plasticity of meningeal lymphatic networks and suggest that compensatory lymphangiogenesis is a major mechanism by which MLVs respond to injury and disease states. This underscores the need for further investigation into how heightened lymphangiogenesis with injury or disease corresponds to functional changes in MLV drainage.

Fig. 2.

Animal models of injury and disease reveal changes to meningeal lymphangiogenesis. A TBI in mice causes a transient rise in the levels of ICP. These animals display increased meningeal lymphangiogenesis 1-week post-injury, which coincides with improved lymphatic drainage. B In a conditional Twist1 knock-out model of craniosynostosis, the cranial sutures fuse prematurely, causing abnormal skull development, which is also predicted to cause elevated levels of ICP. At P16, these mice display hypoplastic dorsal MLVs with reduced coverage and sprouting along the venous sinuses. C Congenital PCLɣ2^−/− mice have reduced lymph flow as a result of blood-filled lymphatics. These animals have hypoplastic meningeal lymphatic vessels at P21, suggesting that loss of laminar flow perturbs the growth and expansion of these networks

Table 1.

MLVs exhibit varied responses to pathological states, and their manipulation produces differing outcomes

Animal model	Lymphatic response (structural)	Timeline of structural response	Associated functions response	Strategy used to modify MLVs	Animal response to MLV changes
Traumatic brain injury	Heightened lymphangiogenesis	1 week post-TBI	Improved CSF uptake	Vegf-c overexpression	Improved outcome
CNS viral infection	Heightened lymphangiogenesis	3 days post-infection	Impaired MLV F\function	Vegf-c treatment	Improved outcome
Alzheimer's disease	None found	N/A	N/A	Vegf-c treatment	Improved outcome
Stroke (focal cerebral ischemia)	None studied	N/A	N/A	Vegfr3 blockade	Improved outcome

In TBI models, compensatory lymphangiogenesis 1-week post-TBI is associated with improved CSF uptake. By contrast, heightened expansion after viral infection occurs within 3 days post-infection and corresponds with impaired MLV drainage of macromolecules. Additionally, augmenting MLVs through Vegf-c treatment or overexpression leads to improved functional and/or behavioral outcomes in TBI, CNS viral infection, and Alzheimer’s models. In a focal cerebral ischemia model of stroke, however, impeding MLV growth via Vegfr3 blockade leads to improved outcomes, suggesting that augmenting MLVs may cause further harm in this model. Together, these results reinforce the dichotomous structural and functional responses of MLVs to changes in their environment and the various therapeutic strategies that can be used to target MLVs in injury and disease

In addition to characterizing how lymphatic networks respond to pathological states, multiple studies have also shown that impaired and/or augmented meningeal lymphatic networks can affect outcomes from CNS injury and disease. In Vegfr3^wt/null mice with hypoplastic meningeal lymphatic networks, infarct volumes after tMCAo are elevated in comparison with sham-treated mice [43]. Similarly, MLV ablation via visudyne/photoconversion in a mouse model of subarachnoid hemorrhage exacerbates neuroinflammation and impairs performance on behavioral tasks [46]. This method of MLV ablation also leads to upregulation of complement-related genes, increased gliosis, and worsened neurologic deficits after TBI [44]. Finally, visudyne photoablation of MLVs in 5xFAD mouse models of Alzheimer’s with intrinsic amyloid pathology causes increased plaque aggregation in the brain and meninges when compared to 5xFAD animals with functional MLVs [42]. Interestingly, in both subarachnoid hemorrhage and TBI, overexpression of Vegf-c via adenoviral therapy, which promotes lymphangiogenesis and helps improve functional drainage, ameliorates these worsened outcomes [44, 46]. Similar results are also seen in aged mice [42]. In addition, Vegf-c adenoviral therapy in 5xFAD mice can improve amyloid-beta clearance following monoclonal antibody treatment [47] (Table 1). However, in a focal cerebral ischemia model of stroke, targeting Vegfr3 signaling to block MLV function improves outcomes by ameliorating systemic inflammation and immune penetration into the brain [48] (Table 1). This suggests that augmenting MLV function may sometimes have unintended neuroimmune effects and can be deleterious in certain pathological states. Still, these results, as a whole, show that further investigation is required into leveraging meningeal lymphatic networks as a therapeutic target to improve outcomes from CNS injury and disease, particularly in the aging population.

The effects of the unique meningeal environment on MLVs

A key feature differentiating MLVs from peripheral lymphatic networks is the specialized environment in which they are housed—within the meninges and at the intersection of the skull and brain. As such, the development, functions, and long-term maintenance of these vessels are affected by their unique meningeal environment, as described above in injury and disease models. A major variable within the meningeal compartment is intracranial pressure (ICP), the maintenance of which is important for fluid dynamics and blood flow. During human infancy, or the first few postnatal weeks in mice, increasing volumes of blood and CSF cause a rapid rise in ICP [49]. Concurrent with this rise in ICP, MLVs grow along the venous sinuses in a basal-to-dorsal manner, from the base to the top of the skull where the confluence of sinuses resides [1] (Fig. 3). Considering ICP varies throughout development and in response to injury, it is important to understand how changes to ICP affect meningeal lymphatic networks and fluid drainage.

Fig. 3.

Developmental timeline of the skull, brain waste clearance pathways, and ICP normalization. During the first postnatal weeks in mice, skull growth (top row) and ICP normalization coincide with the development of MLVs (2nd row) and the onset of brain-CSF perfusion (i.e., the glymphatic system, 3rd row). At P1, the fontanelles of the mouse skull are open, MLVs have not yet reached the dorsal venous sinuses, and CSF circulation and brain influx pathways are beginning to form within the perivascular spaces of pial arteries at the base of the skull. The fontanelles subsequently close (but the cranial suture remain open) and the skull continues to grow as ICP rises (bottom row). During this time, MLVs develop in a basal-to-dorsal manner, reaching the transverse sinuses by P8, the confluence of sinuses by P16, and halfway along the superior sagittal sinus by P20. The growth of MLVs along the venous sinuses is spatiotemporally similar to the basal-dorsal maturation of brain–CSF perfusion pathways along the cerebral arteries, indicated by red shading

In addition to TBI, acute changes to ICP have been modeled in mice via bilateral jugular vein ligation (JVL), causing a steep but transient rise in ICP which normalizes after 24 h [44]. In this model, decreased drainage of CSF macromolecules to the dCLNs was observed, indicating deficits in meningeal lymphatic function [44]. Although morphological changes to MLVs were not assessed in the JVL model, the procedure was performed to mirror the rise in ICP following TBI. As compensatory lymphangiogenesis was observed in TBI models (Fig. 2a, Table 1), it is possible that similar morphological changes may be present in various animal models with acutely elevated ICP [44].

The effects of elevated ICP on MLVs have also been examined in a separate model by inserting and inflating a catheter balloon in the epidural space [50]. In this model, ICP could be more precisely controlled by inflating the balloon to induce varying levels of pressure. Animals with elevated ICP via this method had decreased CSF perfusion into the brain, decreased CSF uptake and drainage by MLVs, and increased spinal lymphatic drainage to the sacral lymph nodes [50]. This study not only reinforced the deleterious effects of elevated ICP on meningeal lymphatic drainage, but also demonstrated that cranial CSF flow pathways are sensitive to and may be altered by variations in the level of ICP, now draining preferentially to spinal lymphatics. In both aforementioned models, however, ICP elevation was transient, and the impact of chronically elevated ICP on meningeal lymphatic networks has yet to be studied in detail. Further, the effects of congenitally raised ICP on the growth and expansion of MLVs is of particular interest due to the temporal overlap between MLV development and ICP normalization.

Early postnatal life is critical not only for MLV development, but also for growth of the skull. Within the first year of life in humans, the neurocranium reaches 65% of its adult size and the brain nearly triples in volume [49, 51]. Due to their coincidental development, it is likely that skull growth, along with ICP normalization, must be tightly regulated to ensure proper development of MLVs and the onset of CSF flow and fluid drainage from the head (Fig. 3). In craniosynostosis, a relatively common craniofacial disorder characterized by premature fusion of the cranial sutures, normal skull growth is disrupted. This can lead to elevated ICP and altered fluid dynamics in the head [52, 53]. Our group has studied the growth and development of MLVs in a Twist1^FLX/FLX:Sm22a-Cre mouse model of craniosynostosis. These mice show reduced sprouting and coverage of dorsal lymphatic vessels at P16, during which time MLVs are actively sprouting and remodeling along the dural venous sinuses [39] (Fig. 2b). Interestingly, effects on MLVs were maintained into adulthood, as these mice still exhibited hypoplastic dorsal MLVs, reduced drainage to dCLNs, and variably affected basal lymphatics, which were often hyperplastic [39]. This suggests that dorsal and basal lymphatics react differently in response to changes in ICP and/or their environment, and that early disruptions to the growth and expansion of MLVs in craniosynostosis continue to affect these vessels later in life. It is notable, however, that Twist1^FLX/FLX:Sm22a-Cre mice also have dural venous sinus malformations and hypoplastic dura, which may have impinged upon Vegfc–Vegfr3 signaling and/or MLV growth along the dural extracellular matrix [39]. Furthermore, the levels of ICP still need to be measured in this model; however, elevated ICP is present in Twist1 heterozygous null mice and human studies suggest that elevated ICP is common in syndromic craniosynostosis [53, 54]. Going forward, it will be important to leverage other models of craniosynostosis, in which the dural venous sinuses and dural extracellular matrix are unaffected, to gain novel insights into the specific, long-term effects of chronically elevated ICP on the development and maintenance of MLVs. This will potentially shed light on how pressure, and potentially its downstream effects on flow dynamics, affect meningeal lymphatic growth and maturation.

Potential role of CSF as a mediator of meningeal lymphangiogenesis

Throughout the circulatory system, mechanical forces generated by circulating fluids and lymph flow are essential for the growth and sprouting of blood and lymphatic vessels [55–57]. Blood flow generates shear stress and circumferential stretch forces necessary for angiogenic sprouting, pruning, lumen formation, and cell fate transitions via direct interactions on the endothelium [55]. Similarly, lymphangiogenesis in peripheral tissues relies upon laminar flow of interstitial fluid (ISF), which directs growth of lymphatic vessels. Seminal work in mouse injury models enabling observation of functional fluid channels in the tail showed that the channeling of ISF precedes the organization of lymphatic networks. Lymphatic vessel growth occurs in the direction of lymph flow, implying that ISF flow helps direct peripheral lymphangiogenesis [56]. MLVs are likewise predicted to be under the influence of laminar flow. In addition to lymph flow in dura, they receive indirect exposure to CSF via mechanisms (or pathways through the arachnoid barrier) that are still not well understood [58]. Moreover, meningeal lymphangiogenesis also mirrors the spatiotemporal onset of preferred CSF circulation routes in the head [59] (Fig. 3). These observations suggest that exposure to ISF and/or CSF may help guide the growth and expansion of meningeal lymphatic networks.

A permissive role of laminar flow for meningeal lymphangiogenesis was first investigated by leveraging a genetic model with reduced lymph flow. In PLCɣ2^−/− animals, there is backflow of blood from the venous system into the lymphatic vasculature, resulting in blood-filled lymphatics with impaired function [60]. Although it was expected that macromolecule clearance to dCLNs would be reduced in PLCɣ2-deficient mice, meningeal lymphangiogenesis was also impaired as evidenced by hypoplastic vessels and less overall coverage at P21, by which time MLV development is normally nearing completion [60] (Fig. 2c). These observations suggest that similar to peripheral lymphatics, postnatal growth and/or remodeling of MLVs is regulated by and sensitive to changes in laminar flow.

Laminar flow may mediate MLV growth and expansion through mechanosensitive ion channels. Piezo1-mediated mechanotransduction signaling facilitates the growth and maintenance of peripheral lymphatic vessels in response to laminar flow, and deletion of Piezo1 causes vessel regression in adult mice [57]. Additionally, administration of Yoda1, a small molecule agonist for Piezo1 that can cross the blood–brain barrier and lowers the threshold for channel activation, has been reported to increase vessel density in lymphatic cultures [57]. In response to shear stress, Piezo1-mediated signaling is also necessary for functional fluid drainage and the formation of lymphatic valves, which is impaired in mouse models lacking Piezo1 in LECs [61]. Furthermore, deletion of Piezo1 in adult mice causes degeneration of existing lymphatic valves, indicating a continued reliance on Piezo1-related signaling for structural maintenance of lymphatic vasculature [62]. Finally, loss-of-function mutations in human Piezo1 are associated with primary lymphedema [63]. Taken collectively, these findings suggest that meningeal lymphangiogenesis may similarly be under the control of Piezo-mediated mechanotransduction signaling, via forces exerted by interstitial flow in dura and/or CSF.

It is also possible that interstitial flow regulates meningeal lymphangiogenesis via remodeling of the extracellular matrix and integrin-dependent activation of Vegfr3 and/or by providing MLVs with access to Vegf-c. As interstitial fluid builds during embryogenesis concomitant with the growth of blood vessels, increased volumes of ISF activate integrin-B1 receptors in the extracellular matrix. These receptors associate with and phosphorylate Vegfr3, activating downstream signaling and stimulating the proliferation of lymphatic endothelial cells in a manner independent from activation by Vegf-c [64]. It is also proposed that mechanostimulation disrupts interactions between integrin-B1 receptors and integrin-linked kinase, thus enabling integrin-B1 to interact with Vegfr3 [65]. Additionally, CSF may act as a vehicle to deliver Vegf-c to meningeal LECs, thereby promoting Vegfc–Vegfr3 signaling in a flow-dependent manner. In one study utilizing mouse models of glioblastoma, administration of soluble Vegfr3 into the CSF to sequester Vegf-c mitigated the efficacy of Vegf-c mRNA, suggesting that Vegf-c may utilize CSF flow as a conduit to reach and act upon meningeal LECs [66]. As such, it is tempting to speculate that CSF may help mediate growth factor-dependent lymphangiogenesis, while also providing flow-dependent regulation of lymphangiogenesis via integrin-mediated activation of Vegfr3.

Accumulating evidence has shown that laminar ISF flow is integral for the growth and expansion of peripheral lymphatic networks, but more evidence is needed to understand how flow and mechanical forces shape the growth, expansion, and even long-term maintenance of MLVs. Investigating how fluid forces (especially those exerted by CSF) shape the development and maintenance of MLVs is important, because CSF flow and turnover naturally declines with age [67]. Importantly, aging is associated with pathological changes to MLVs, including regression of dorsal networks and hyperplastic basal vessels [42]. As discussed above, TBI models with elevated ICP and reduced CSF drainage to the dCLNs are also associated with pathological changes to MLVs [44, 50]. Thus, establishing the functional implications of CSF on meningeal lymphangiogenesis and pathological changes in the absence of proper flow is likely a critical therapeutic target in aging and MLV-associated pathology.

Given the various origins and functions of MLVs, it is also important to investigate how mechanical forces work alongside other factors such as ICP, skull growth, and transcriptional regulation to facilitate their growth and development. MLVs along the middle meningeal artery, for example, do not appear to have to access to CSF, as they are not labeled by fluorescently-conjugated antibodies injected i.c.m. [2]. Given this finding, it is likely that the function of these vessels differs regarding CSF drainage and that laminar flow exerted by CSF may play less of a role in their development. This also raises important questions, such as: Are perisinusoidal lymphatics under differential developmental control compared to MLVs adjacent to arteries? Are perisinusoidal lymphatics more specialized for CSF uptake versus arterial lymphatics, which may play a larger role in draining dural ISF? How do the cellular origins of lymphatic populations predict the factors that govern their growth and long-term maintenance? What insights can we glean from the development of these vessels regarding their key functions and responses to pathological states? It is the answers to these questions—and more—that will shape our understanding of meningeal lymphatics and pave the way for targeted treatments for a host of developmental, neurodegenerative, and craniofacial disorders.

Conclusion

The cellular and molecular mechanisms that guide meningeal lymphangiogenesis appear to be controlled by several factors, some of which are shared with peripheral lymphatics (i.e., Vegf-c/Vegfr3 signaling) and others which are less understood and unique to these vessels’ meningeal environment (i.e., ICP and CSF flow). Refining our understanding of how these factors and processes coalesce will reveal important insights into meningeal lymphangiogenesis and maintenance across the lifespan, as well as functional decline in aging and disease. By revealing the similarities and unique aspects of lymphangiogenic regulation in normal development compared to injury and disease response, we can better understand how to leverage meningeal lymphatics as a therapeutic target via adenoviral therapies and small molecule approaches. Furthermore, understanding how MLVs integrate with brain waste clearance systems (i.e., glymphatic system) during development and throughout the lifespan is particularly important for understanding the complete pathway by which toxic and metabolic waste is cleared from the CNS. Ultimately, given the critical role of waste clearance for preventing neurodegenerative disease and repair following injury, further study into the development and maintenance of MLVs in animal models of injury and disease holds promise for harnessing the novel therapeutic potential of this system.

Author contributions

The first draft of the manuscript was written by (AJ and PSA), with review and editing by (MJM and MAT)

Funding

This work was supported by NIDCR 1R03DE032409-01 (MAT).

Data availability

Enquiries about data availability should be directed to the authors.

Declarations

Conflict of interest

All authors declare that they have conflict of interest.

Consent for publication

Not applicable.

Ethical approval and consent to participate

Not applicable.