Vasomotor influences on glymphatic-lymphatic coupling and solute trafficking in the central nervous system

Abstract

Despite the recent description of meningeal lymphatic vessels draining solutes from the brain interstitium and cerebrospinal fluid (CSF), the physiological factors governing cranial lymphatic efflux remain largely unexplored. In agreement with recent findings, cervical lymphatic drainage of 70 kD and 2000 kD fluorescent tracers injected into the adult mouse cortex was significantly impaired in the anesthetized compared to waking animals (tracer distribution across 2.1 ± 4.5% and 23.7 ± 15.8% of deep cervical lymph nodes, respectively); however, free-breathing anesthetized mice were markedly hypercapnic and acidemic (paCO₂ = 64 ± 8 mmHg; pH = 7.22 ± 0.05). Mechanical ventilation normalized arterial blood gases in anesthetized animals, and rescued lymphatic efflux of interstitial solutes in anesthetized mice. Experimental hypercapnia blocked cervical lymphatic efflux of intraparenchymal tracers. When tracers were injected into the subarachnoid CSF compartment, glymphatic influx into brain tissue was virtually abolished by hypercapnia, while lymphatic drainage was not appreciably altered. These findings demonstrate that cervical lymphatic drainage of interstitial solutes is, in part, regulated by upstream changes in glymphatic CSF-interstitial fluid exchange. Further, they suggest that maintaining physiological blood gas values in studies of glymphatic exchange and meningeal lymphatic drainage may be critical to defining the physiological regulation of these processes.

Introduction

In addition to classically defined routes of subarachnoid cerebrospinal fluid (CSF) reabsorption into the dural sinuses via arachnoid granulations, many reports, including work from the Westrop, Cserr, and Johnston laboratories, have demonstrated that CSF efflux also occurs via lymphatic reabsorption.^1–7 Indeed, Bradbury and Westrop showed that a major route of lymphatic CSF efflux occurs at the cribriform plate in the base of the skull using cyanoacrylate glue to seal this pathway.³ More recently, interest in this pathway has been reinvigorated by the description of a lymphatic capillary network in the meninges of rodents^8,9 and humans.^10,11 To date, meningeal lymphatic vessels have been most extensively characterized in the calvarium,^9,12 although lymphatic capillaries also exist in the skull base.⁸ In early reports considering meningeal lymphatic function, some studies have reported direct solute uptake by these lymphatic capillaries in the calvarium,⁹ while others have proposed that the majority of lymphatic solute efflux occurs in the base of the skull.^13,14 While the relative importance of these sites in lymphatic solute uptake remains to be established, even less is known about physiological regulators of this efflux pathway.

In peripheral organs, lymphatic reabsorption of solutes in the interstitial fluid (ISF) is mediated by cyclical changes in the extracellular volume, a process termed extrinsic pumping. Under anesthesia, peripheral lymphatic clearance of ISF solutes is slowed and is rescued by physical compression of the tissue.¹⁵ However, the unique anatomical environment of the intracranial space may confer unique properties to lymphatic drainage from the brain ISF. The physiological and pathophysiological processes that govern solute exchange between the brain ISF, CSF and lymphatic compartments, and their relationships to one another are subjects of considerable current interest. For example, the exchange of tracers between the CSF and ISF compartments along the perivascular pathways that comprise the glymphatic system^16,17 is driven by arterial pulsation^18,19 and is more rapid in the sleeping compared to the waking state.^20,21 A recent study reported that lymphatic drainage from the central nervous system (CNS) is markedly reduced in anesthetized mice compared to awake mice, suggesting that arousal state affects lymphatic solute drainage.¹³ Based on these observations, the authors concluded that more rapid lymphatic clearance of CSF tracer in the waking state may underlie the previously observed slowing of CSF-ISF tracer exchange in awake mice.²⁰

While these studies raise important questions about solute exchange between the ISF, CSF and lymphatic compartments, many physiological changes associated with anesthesia may account for these discrepant findings. Specifically, anesthesia can dramatically alter an animal's physiological state by reducing its respiratory drive, leading to derangements in arterial blood gas (ABG) values.²² Therefore, we first sought to replicate the previously reported effects of ketamine-xylazine anesthesia on lymphatic drainage of brain interstitial tracers. Using mechanical ventilation and acute hypercapnic challenge, we then defined the effect that hypercapnic vasodilation has on both lymphatic drainage of interstitial solutes, as well as the exchange of CSF tracers into the interstitial and lymphatic compartments.

Materials and methods

Animals

Male and female 8- to 12-week-old C57BL/6 mice from Jackson Laboratories were used for experiments. Mice were acclimated to their housing for several days before experiments were performed. For all surgical experiments, the core body temperature was monitored via rectal thermometer and maintained using either a heating pad or heating lamp and mice were weighed to determine dose of anesthetic (ranging from 15 to 25 g). All experiments were approved by the Institutional Animal Care and Use Committee of Oregon Health & Science University. The nature of the procedures performed precluded groupwise blinding during surgeries; however, all following analyses were performed with groupwise blinding. These experiments were conducted according to the Guide for the Care and Use of Laboratory Animals, established by the Association for Assessment and Accreditation of Laboratory Animal Care. The results described in this manuscript are compliant with the Animal Research: Reporting in Vivo Experiments (ARRIVE) guidelines.

Intraparenchymal tracer injections

Anesthesia was induced using 3% isoflurane and animals were placed in three-point stereotaxic apparatus. Isoflurane was reduced to 2% and a surgical plane of anesthesia was ensured throughout the procedure by regular pedal reflex checks. A 2 cm midline incision was made in the scalp over the sagittal suture and the skin was reflected to locate bregma. After allowing the skull to dry, a burr hole was introduced in the right parietal bone (0.4 mm caudal, 1 mm lateral to bregma) using a Microtorque II drill. A 26 Ga needle (Hamilton) was inserted to a depth of 1 mm and rested for 10 min before injecting 1 µL of 1:1 solution of 70 and 2000 kDa dextran in normal saline (25 mg/mL and 10 mg/mL, respectively; ThermoFisher) at a rate of 100 nL/min. During the injection, for animals that were to remain under anesthesia, isoflurane was tapered off and mice received an intraperitoneal dose of ketamine/xylazine (1.75 mg ketamine/0.25 mg xylazine per 20 gram mouse). The needle was removed from the brain following the injection and incision was closed with cyanoacrylate glue. Animals were removed from the stereotaxic apparatus and either placed on a heating pad for the duration of the experiment (anesthetized group) or placed in their housing box after righting response was observed (awake group). Animals were sacrificed at 30, 60, and 120 min via intracardiac infusion of heparinized saline followed 4% paraformaldehyde. In the event that tracer was injected into the lateral ventricle, animals were excluded from the study (n=4).

Intracisternal tracer injections

For intracisternal injections, animals were placed in three-point stereotaxic apparatus. A 2 cm midline incision was made caudally from lambda. Skin was reflected laterally and blunt dissected to expose muscle. The most superficial muscle layer was sectioned and reflected laterally, the deeper layers were kept intact, and retracted laterally using 6-0 braided suture. The atlanto-occipital membrane was visualized via blunt dissection through ligamentous tissue and the bevel of a 30.5-gauge needle attached to PE-10 tubing was inserted into the cisterna magna. 2 µL 1:1 solution of 70 and 2000 kDa dextran in normal saline (25 mg/mL and 10 mg/mL, respectively; ThermoFisher) was infused into the cisterna magna at a rate of 1 µL/min. During the injection, for animals that were to remain under anesthesia, isoflurane was tapered off and mice received an intraperitoneal dose of ketamine/xylazine (1.75 mg ketamine/0.25 mg xylazine per 20 g mouse). To minimize tracer leakage, cyanoacrylate glue was placed over the atlanto-occipital membrane and the incision was reapproximated. All mice that were awakened displayed the righting reflex within 8 minutes of the injection start time. Thirty minutes after initiation of the injection, animals were fixation perfused. In the event that blood was observed in the subarachnoid space after perfusion, animals were excluded from the study (n=2).

Mechanical ventilation

For experiments featuring mechanical ventilation, mice were anesthetized with 3% isoflurane and endotracheally intubated with a Becton Dickinson intravenous 22 ga catheter after Rivera et al.²³ Mice were ventilated immediately on 100% inspired O₂ after intubation using a mechanical ventilator designed specifically for mice (Hugo Sachs Elektronik MiniVent type 845). After completion of the tracer injection, mice that were in the “Awake” group were tapered off isoflurane, extubated, and allowed to freely behave in their home cage. Mice in the “300 bpm” and “300 bpm + 5% CO2” received an intraperitoneal dose of ketamine/xylazine and isoflurane was quickly tapered off. In experiments featuring hypercapnic challenge, the inspired gas was switched from 100% O₂ to a mixture of 5% CO₂ and 95% O₂ throughout the duration of hypercapnic challenge. In order to match the timeline of hypercapnia to the timeframe over which lymphatic drainage was being evaluated, we measured the effect of hypercapnia at t = 15 min. In all experiments, the end expiratory pressure was set to +3 cm H₂O, the inspiratory pressure was set to +10 cm H₂O, and tidal volume was set to 250 µL. Symmetrical chest rise and consistent effluent gas in the positive end expiratory column during expiration were used as criteria for accurate placement of the endotracheal tube. After insuring successful intubation, mice were paralyzed with an intraperitoneal dose of succinylcholine (25 mg/kg) to prevent the animals from challenging the ventilator. Heart rate and muscle tone readings were monitored using electrocardiogram and electromyography to insure depth of anesthesia but not recorded for experimental purposes.

Monitoring of physiological parameters

Arterial blood for blood gas measurements was either collected at the end of experiments or repeated sampling from the femoral artery. After termination of the experiment, blood was collected directly from the left ventricle via cardiac puncture. For repeated blood gas measurement, the superficial femoral artery was isolated and ligated with 5-0 braided suture. Immediately proximal to the ligature, a small incision in the arterial wall was made with microvascular scissors, with temporary compression applied proximal to the incision to prevent bleeding. Blood was collected directly from the femoral artery at relevant time points. All arterial blood samples were analyzed immediately using a handheld blood gas analyzer (iStat, CG8 + cartridge). Cerebral blood flow and intracranial pressure were monitored simultaneously during experiments on an independent cohort of mice. To accomplish this, mice were placed in a three-point stereotactic apparatus during mechanical ventilation and the scalp was visualized using a chemical depilatory agent. A 2 cm midline incision was made and the scalp was reflected laterally to visualize the cranium. A Laser Doppler Flometry probe (Moor Instruments DRT4) was placed over the left parietal bone to measure cerebral blood flow. For intracranial pressure (ICP) monitoring, a small burr hole was made in the right parietal bone using a Microtorque II drill. The burr hole was carefully made to avoid damaging the dura mater, and after visualizing the brain, a pressure transducer (SPR-1000, Millar Instruments) was lowered through the dura mater into the parietal cortex for intraparenchymal monitoring. To insure accurate probe placement and monitoring at the beginning and end of every experiment, transient compression of the abdomen was used to approximate a Valsalva maneuver, which transiently increased measured ICP by 25–35 mmHg before rapidly returning to baseline.

Microscopy

Mice were fixation-perfused with fresh 4% paraformaldehyde stored at 4℃. The calvarium, brain, skull base, and deep cervical lymph nodes (DCLNs) were harvested and post-fixed in 4% paraformaldehyde overnight at 4℃. Calvaria, skull bases, and DCLNs were imaged immediately using fluorescence dissection microscopy (Zeiss AxioZoom V16). Brains were dehydrated by submersion in 30% sucrose overnight at 4℃, frozen in OCT by submersion in 2-methylbutane on dry ice, and stored at −80℃. Brain sections were collected as floating sections in PBS. Sections were washed 3 × with PBS and mounted using mowiol 4-88 mounting media. All sections were imaged using either the LSM880 (Zeiss) or AxioScan (Zeiss) within 48 hours of mounting using the same laser power, photomultiplier voltage, and gain across groups. Image post-processing and analysis was performed using ImageJ. This includes image thresholding and subregional brain analysis. All image post-processing was performed blinded to experimental groups.

Statistics

All statistical analyses were performed using GraphPad Prism 8 and assembled in Adobe Illustrator. Data are presented as the mean ± standard deviation in the text. Box and whisker plots shown in figures represent the median (bar), interquartile range (box), and total range (error bars). For groupwise comparisons, one-way or two-way ANOVA with and without repeated measures were implemented, with Sidak's post hoc correction for multiple comparisons. An unpaired t-test with Welch's correction was used for pairwise comparisons. Results with a p value less than or equal to 0.05 are considered significant.

Results

Cervical lymphatic drainage of brain interstitial tracer is slowed in free-breathing anesthetized mice

To evaluate the effect of anesthesia on interstitial solute distribution and lymphatic drainage, we co-injected 70 kDa and 2000 kDa dextrans into the motor cortex of adult mice. Upon completion of the injection, mice were either awakened and allowed to behave freely in their home cage or remained under ketamine/xylazine anesthesia for 30, 60, or 120 min (Figure 1(a) and Supplemental Figure 1A). In mice survived for 120 minutes, we measured fluorescence intensity in subregions from three representative coronal sections to map tracer distribution in the brain parenchyma (Figure 1(b) and Supplemental Figure 1B). We observed that within the cortex, fluorescence intensity of 70 kDa intraparenchymal tracer remained elevated in waking compared to anesthetized animals (Figure 1(b); p = 0.028 ipsilateral, p = 0.054 contralateral; two-way repeated measures ANOVA; n = 5 per group). This effect was not observed in the subcortical white matter of the corpus callosum (Figure 1(b); p = 0.596 ipsilateral, p = 0.368 contralateral). No difference in the distribution of the intraparenchymal 2000 kDa tracer was observed in the cortex or subcortical white matter between kextamine/xylazine-anesthetized compared to waking mice (Supplementary Figure 1B).

Figure 1.

Anesthesia reduces interstitial retention and slows lymphatic drainage of intraparenchymal tracer. (a) Schematic of intraparenchymal tracer injection protocol. (b) Left: Heat map of dextran-70 distribution 120 min following intracortical injection in awake vs. ketamine-xylazine-anesthetized mice (n = 5–6 per group). Black and gray shaded regions indicate cortical and corpus callosum ROIs, respectively. Right: Intraparenchymal tracer retention was significantly greater in the ipsilateral (p<0.05 anesthetized vs. awake, n = 5–6 animals per group) and trended toward greater distribution in the contralateral cortex (p=0.054 anesthetized vs. awake) of waking compared to anesthetized animals. (c) 30 min following intra-cortical tracer injection, tracer associated with the dorsal meninges (M) and superior sagittal sinus (SSS) or the ventral skull base meninges or trigeminal nerves (CN V) did not differ between waking or anesthetized states (n = 5–6 per group). (d) Drainage of intra-cortical tracer to the deep cervical lymph nodes is slowed in anesthetized compared to waking animals (**P_adj=0.006; n = 5–6 per group).

To define the distribution of interstitial solutes within the cranium after they exit the brain, we evaluated fluorescence intensity within the calvarium (dorsal skull surface) and skull base in mice 30 min after intracortical injection of 70 kDa and 2000 kDa fluorescent tracers. Within the calvarium, we measured fluorescence intensity around the superior sagittal sinus (SSS) and the meninges adjacent to the SSS (Figure 1(c) and Supplemental Figure 1C). In the skull base, we measured fluorescence intensity near the trigeminal nerve (CN V) and in the meninges overlying the rostral aspect of the basiphenoid bone (Figure 1(c) and Supplemental Figure 1C). Although we did not observe an effect of anesthesia on the cranial distribution of tracer, we found that the vast majority of tracer was distributed along the base of the skull, associated with the meninges and cranial nerves, compared to the dorsal calvarium. This was true of both 70 kDa and 2000 kDa intraparenchymal tracers.

To define the effect of anesthesia on lymphatic clearance of interstitial solutes in free-breathing animals, we harvested and measured fluorescence intensity of tracer in the DCLNs from mice at 30, 60, and 120 min following injection (Figure 1(d) and Supplemental Figure 1D). Interestingly, we observed a significant reduction in the rate of lymphatic tracer clearance in anesthetized animals compared to awake animals (70 kD tracer p = 0.0061, 2000 kD tracer p = 0.0018; two-way ANOVA; n = 5–6 per group). We also observed that although the dextran tracers were injected into the right motor cortex, the tracer distribution occurred bilaterally in the DCLNs, suggesting that lymphatic drainage of interstitial solutes is not lateralized (Supplemental Figure 2).

Hypercapnia in free-breathing anesthetized mice is normalized by mechanical ventilation

Several types of anesthesia, including ketamine/xylazine are known to depress the hypercapnic ventilatory response, which is responsible for modulating respiratory rate in response to changes in the blood pH.²² This results in the development of an acute respiratory acidosis, characterized by reduced respiratory rate, lowered blood pH, increased arterial tension of CO₂, and normal HCO₃⁻. To determine the effect of ketamine/xylazine anesthesia on the ABGs of wild type mice, we anesthetized mice and sampled blood from the superficial femoral artery after 30 minutes of anesthesia. We observed a significant respiratory acidosis in non-intubated free-breathing mice, with a blood pH of 7.22 ± 0.05, and paCO₂ of 64 ± 8 mmHg (Table 1). Mice were also hypoxemic and oxygen desaturated, with an average paO₂ and oxygen saturation of 50 ± 8 mmHg and 74 ± 9%, respectively. To correct these derangements in blood gases, we endotracheally intubated mice and empirically determined ventilation settings that corrected the ABGs. We found that mice ventilated at a rate of 300 breaths min⁻¹ (bpm) on either oxygen-supplemented room air (data not shown) or 100% O₂ corrected the paCO₂, paO₂, O₂ saturation, and pH (Table 1). When anesthetized animals were subjected to a 30 minutes hypercapnic challenge at a respiratory rate of 300 bpm, paCO₂ levels increased from 43 ± 1 mmHg under normocapnic conditions to 66 ± 5 mmHg. Similarly, blood pH declined from 7.38 ± 0.04 to 7.18 ± 0.03 with hypercapnic challenge. O₂ saturation, PaO₂ and HCO₃ levels were not altered by hypercapnic challenge (Table 1).

Table 1.

Characterization of arterial blood gases in non-intubated, intubated, and CO₂-challenged mice.

	Free breathing (147–169 bpm)	200 bpm	250 bpm	300 bpm	300 bpm + 5% CO2
Oxygen saturation (%)	74 ± 9	100 ± 0	100 ± 0	100 ± 0	100 ± 0
PaO2	50 ± 8	452 ± 19	455 ± 14	431 ± 26	454 ± 22
PaCO2	64 ± 8	70 ± 7	58 ± 1	43 ± 1	66 ± 5
HCO3	26 ± 0.3	26 ± 2	27 ± 2	25 ± 3	25 ± 1
pH	7.22 ± 0.05	7.18 ± 0.06	7.27 ± 0.04	7.38 ± 0.04	7.18 ± 0.05

Arterial blood gas measurements in non-intubated mice and intubated mice mechanically ventilated at different respiratory rates.

Intracranial distribution and drainage of parenchymal solutes in mechanically ventilated mice

We surmised that the substantial hypercapnia and acidosis observed in freely breathing anesthetized mice may contribute to the observed differences in interstitial tracer efflux along the cervical lymphatic pathway (Figure 1(c) and (d)). In both humans and rodents, hypercapnia evokes vasodilation and increased cerebral blood flow,^24–26 which may alter the exchange of CSF and interstitial solutes along perivascular spaces. Similarly, perivascular glymphatic exchange may slow in response to the reduced vascular pulsatility observed during hypercapnia.^19,27 Therefore, we tested the effect of hypercapnic challenge on the lymphatic drainage of interstitial tracers, comparing anesthetized mice that were mechanically ventilated at 300 bpm with or without exposure to 5% CO₂. These groups were subsequently compared to animals that were awake and freely behaving following intraparenchymal tracer injection. Mice in all groups were sacrificed via perfusion fixation 30 minutes after the injection was initiated, and tracer intensity in the calvaria, brains, skull bases, and DCLNs of these mice was measured to estimate tracer distribution in each compartment.

Similar to the findings in awake and nonintubated anesthetized mice at 30 minutes, we did not observe any difference in the parenchymal distribution of tracers among ventilation states (data not shown). In these animals, we again noted greater tracer distribution to the ventral skull base compared to the dorsal calvarium (Figure 2(a), Supplementary Figure 3A). Interestingly, anesthetized animals undergoing normocapnic ventilation exhibited significantly greater distribution to the dorsal calvarium compared to awake animals (p = 0.0113, one-way ANOVA, n = 7 per group). Normocapnic ventilation in anesthetized animals resulted in interstitial tracer drainage to the DCLNs that was not statistically different than that observed in waking animals (Figure 2(b); p = 0.1181; one-way ANOVA; n = 6–7 per group). Hypercapnic ventilation virtually abolished the lymphatic drainage of interstitial tracers in anesthetized animals (Figure 2(b), Supplementary Figure 3B).

Figure 2.

Intracranial tracer distribution and lymphatic drainage of parenchymal tracer in mechanically ventilated mice. (a) 30 min after intracortical injection, tracer levels associated with the SSS (*P_adj < 0.05, n = 6–7 per group) were greater among mechanically ventilated (300 bpm) KX-anesthetized mice compared to waking mice. Association of tracer with dorsal structures was reduced with hypercapnic ventilation (P_adj = 0.17). (b) Intracortical tracer drainage to the DCLNs was not significantly different between waking and ventilated KX-anesthetized states (P_adj = 0.1181). Hypercapnic ventilation abolished lymphatic tracer drainage (*P_adj = 0.0311, ***P_adj < 0.001, n = 6–7 per group).

Intracranial distribution and drainage of CSF solutes in mechanically ventilated mice

Prior studies have demonstrated that ketamine/xylazine anesthesia accelerates CSF tracer influx²⁰ into and interstitial solute efflux^20,28 from the brain compared to the waking state. A recent study also reported that CSF tracer infused into the ventricles re-entered brain tissue more rapidly under ketamine + meditomidine anesthesia than in the waking state.¹³ In this latter study, an inverse relationship between exchange of CSF tracer into the brain interstitium and its clearance along the cervical lymphatic drainage was reported in anesthetized and waking mice, with more rapid lymphatic clearance and slowed parenchymal penetration of CSF tracer in awake animals.¹³ To define whether the cervical lymphatic clearance of interstitial solutes is influenced by the dynamics of glymphatic CSF-ISF exchange, we evaluated whether the movement of subarachnoid CSF tracer into brain tissue versus its direct cervical lymphatic drainage was impaired by experimental hypercapnia. To test this, we co-infused 70 and 2000 kDa dextrans directly into the CSF at the cisterna magna of mice that were ventilated at 300 bpm, 300 bpm + 5% CO₂, or 300 bpm and subsequently awakened. For both 70 kDa and 2000 kDa CSF tracers, influx into the parenchyma was significantly reduced in awake compared to mechanically ventilated anesthetized animals across the rosto-caudal axis (Figure 3(a) to (c) and Supplemental Figure 4A–C; p = 0.0003 and p = 0.0011, respectively; two-way repeated measures ANOVA; n = 8–9 per group). Hypercapnic ventilation of anesthetized animals reduced CSF tracer influx to levels observed in awake mice.

Figure 3.

Hypercapnia impairs CSF-ISF exchange in KX-anesthetized mice. (a) Representative coronal brain sections from mice fixed 30 min following intracisternal CSF tracer (dextran-70) injection. (b) Whole slice tracer coverage was significantly greater in mechanically ventilated (300 bpm) KX-anesthetized compared to awake mice (***P_adj < 0.001, n = 6–10 animals per group). Hypercapnic ventilation significantly reduced CSF tracer influx in KX-anesthetized animals (**P_adj = 0.005; n = 6–10 animals per group). (c) Analysis of different brain regions demonstrates that CSF tracer coverage was significantly reduced across the total slice, and cerebral cortex in waking compared to mechanically ventilated KX-anesthetized mice (****P_adj < 0.001, n = 8–10 animals per group). Hypercapnic ventilation significantly reduced CSF tracer influx in KX-anesthetized animals.

Following intracisternal injection of CSF tracer, no significant differences in tracer distribution to the dorsal calvarium or the ventral skull base were observed (Figure 4(a); p = 0.051 and p = 0.189, respectively and Supplemental Figure 5; p = 0.072 and p = 0.201, respectively; one-way ANOVA; n = 9–10 per group). In contrast to the cervical lymphatic drainage of intraparenchymal tracer, no significant differences in lymphatic drainage of intracisternal CSF tracer were observed among mechanically ventilated anesthetized mice, awake mice, and anesthetized mice subjected to hypercapnic ventilation (Figure 4(b) and Supplemental Figure 5; p = 0.328 and p = 0.218, respectively; one-way ANOVA; n = 9–10 per group).

Figure 4.

Hypercapnia does not alter lymphatic drainage of CSF tracers. (a) 30 min after intracisternal injection, CSF tracer association with dorsal and ventral structures did not demonstrate groupwise differences among waking, hypercapnic, or normocapnic anesthetized mice (n = 9–10 per group). (b) Lymphatic drainage of CSF tracer was not significantly different between mechanically ventilated KX-anesthetized and waking mice (p = 0.1177), nor with hypercapnic ventilation (p = 0.9226, n = 9–10 animals per group).

Monitoring cerebral blood flow and intracranial pressure during hypercapnic challenge

To examine possible mechanisms underlying the reduced influx of CSF tracers into the brain parenchyma and lymphatic drainage of interstitial solutes in response to hypercapnia, we monitored changes in cerebral blood flow and intracranial pressure in an independent cohort of animals. Values were measured at baseline in anesthetized mechanically ventilated (300 bpm) mice, and for 15 minutes following initiation of hypercapnic challenge (300 bpm + 5% CO₂). At the end of each experiment, we confirmed the presence of hypercapnia and acidemia via ABG measurement (paCO₂ = 81.6 ± 7.2 mmHg, pH = 7.163 ± 0.06). Although we observed a modest increase in cerebral blood flow after induction of hypercapnia (117% ± 20% of baseline), no significant change in intracranial pressure was observed (91% ± 18%).

Discussion

The observation that the perivascular exchange of CSF and ISF, termed ‘glymphatic’ function, is a feature primarily of the sleeping brain has attracted much attention since the initial publication in 2013.^20,29 In this study, the influx of CSF tracers into the cortex and the efflux of ISF tracers out of the brain were slowed in awake mice compared to either sleeping or ketamine/xylazine-anesthetized mice. Although one study has failed to replicate the observation that glymphatic exchange is more rapid in animals anesthetized with ketamine/xylazine compared to awake animals,³⁰ a recent study by Ma et al. reported that fluorescent tracer infused into the lateral ventricle entered the brain interstitium more rapidly under ketamine/medetomidine anesthesia than under waking conditions.¹³ In the same animals, reduced cervical lymphatic drainage of CSF tracer was observed in the anesthetized state. From these findings, the authors concluded that the rapid clearance of CSF tracers along lymphatic pathways during waking prevented the influx of CSF tracer into the brain interstitium, accounting for apparent sleep-wake differences in CSF-ISF tracer exchange.

In the present study, we report similar findings to those of Ma et al.,¹³ observing that cervical lymphatic efflux of intracortically injected interstitial tracers was more rapid in waking than in ketamine/xylazine-anesthetized mice. However, we further observed that free-breathing animals under these anesthesia conditions exhibited low respiratory rates and were markedly hypercapnic and acidemic. When arterial paCO₂ and blood pH were normalized by mechanical ventilation, the differences in lymphatic efflux of interstitial tracer between the anesthetized and waking state were no longer present. Confirming that hypercapnia of this magnitude can suppress lymphatic efflux of interstitial solutes, ventilation of anesthetized animals with a 95% O₂ + 5% CO₂ gas mixture virtually abolished lymphatic drainage of interstitial tracers. These findings argue that when animals are maintained under anesthetized conditions that preserve physiological ABGs, lymphatic efflux of interstitial solutes is not altered by anesthesia. Because the study by Ma et al. did not employ mechanical ventilation and did not report ABGs, it is possible that the reported results are attributable to respiratory depression and hypercapnia under ketamine/medetomidine anesthesia.

In recent studies, perivascular glymphatic CSF-ISF exchange^16,17,20 and meningeal lymphatic drainage of the CSF and interstitial compartments^8,9 have primarily been described separately. However, it is speculated that these two processes function in concert to support the lymphatic efflux of interstitial solutes not cleared across the blood brain barrier.³¹ Indeed, a recent study reported that photochemical ablation of the meningeal lymphatic vasculature impaired both glymphatic CSF-ISF exchange and meningeal lymphatic drainage of CSF solutes.³² In the present study, we demonstrated that experimental hypercapnia dramatically slowed the cervical lymphatic drainage of intracortically injected tracers. When the effect of this intervention on the distribution of CSF tracer was evaluated, we observed a dramatic slowing of glymphatic CSF-ISF exchange, but no detectable change in the cervical lymphatic drainage of CSF tracers. These findings suggest that glymphatic exchange and meningeal lymphatic drainage are functionally linked under physiological conditions, with modulation of glymphatic CSF-ISF exchange impacting downstream lymphatic drainage of interstitial solutes. We observed that once intra-cortical tracer entered the CSF compartment, its cranial distribution did not differ between awake and anesthetized mice, concentrating under both conditions in the meninges and peri-neural spaces in the ventral skull base. This observation that interstitial solutes from brain tissue drain preferentially into the skull base rather than the dorsal meninges is consistent with a recently published study which demonstrated that meningeal lymphatic absorption of tracers injected into the CSF mainly occurs in the skull base.³³ It is important to note that, while we observed these patterns of intracranial distribution, we evaluated lymphatic drainage at the level of the DCLNs, therefore we cannot definitively conclude whether the tracers appearing in the DCLNs transited across the skull base or the meningeal lymphatic vessels elsewhere in the cranium. Similarly, we cannot confirm whether the eventual clearance of tracers quantified in these regions occurs via direct absorption into the bloodstream or by lymphatic efflux. However, interstitial tracer distributed bilaterally to both DCLNs, despite being injected unilaterally into the cortex. This suggests that while changes in glymphatic CSF-ISF exchange affect the lymphatic clearance of interstitial solutes, cervical lymphatic drainage occurs bilaterally from the CSF compartment.

One mechanism that may contribute to the slowing of glymphatic CSF-ISF exchange in hypercapnic mice is the direct effect of hypercapnia on the cerebral vasculature. Hypercapnic vasodilation, which occurs in rodents and humans,^24–26 may slow perivascular solute exchange as dilation of arteries and arterioles reduces the cross sectional area of CSF-filled perivascular spaces along which CSF-ISF exchange occurs. Hypercapnic vasodilation additionally reduces cerebrovascular pulsatility, as resistance vessels approach their maximal diameters. In the rabbit, hypercapnic challenge resulting in arterial tension of CO₂ similar to that in our study (∼64 mmHg) reduced cerebrovascular pulsatility index by 47 ± 16%.²⁷ In vivo studies of perivascular CSF-ISF exchange demonstrate that arterial pulsatility is a key driver of glymphatic exchange.^18,19 Therefore, reductions in vascular pulsatility due to hypercapnia may underlie the impairment of CSF-ISF exchange that we observed in hypercapnic mice. It is also possible that ICP changes secondary to increased cerebral blood flow during hypercapnia may alter lymphatic drainage of interstitial solutes. However, in the present experimental setting we were unable to detect substantial changes in ICP during hypercapnic challenge, despite detecting increased CBF using transcranial laser Doppler flowmetry. One limitation of our ventilation approach is the use of 100% O₂ in the inspired gas, which may have confounding effects on CBF and ICP, since elevated paO₂ can cause vasoconstriction in cerebral blood vessels. Although the vasodilatory effects of hypercapnia are widely reported to be more potent than the vasoconstrictive effects of hyperoxia, it remains possible that CBF and ICP may be influenced by hyperoxia.

The observation that vasomotor changes can modulate glymphatic CSF-ISF exchange may have important implications for the understanding of the coupling between brain metabolic function and interstitial solute clearance. Cerebral blood flow is tightly coupled to neuronal activity through the mechanisms of neurovascular coupling, with local changes in activity resulting in local capillary and arteriolar dilation.³⁴ In addition to such metabolic coupling, low-frequency (0.1–0.4 Hz) oscillations in cerebrovascular diameter, termed vasomotion, occur spontaneously in the rodent and human brain.^35,36 While the physiological role of vasomotion has remained unclear, our present results suggest that the role of such low-frequency vasomotor changes in modulating perivascular CSF-ISF exchange may be an important direction for future research. Whether age- or disease-related^37,38 changes in vasomotion confer vulnerability to pathological processes such as amyloid plaque deposition is a similarly important subject of future investigations.

More directly, several clinical conditions are associated with transient or chronic periods of hypercapnia. Both obstructive sleep apnea (OSA) and chronic obstructive pulmonary disease are associated with transient and static periods of hypercapnia, as well as impairments in attention, memory and executive function.^39–41 Recent findings in subjects with OSA suggest that in addition to sleep fragmentation and hypoxemia, hypercapnia may be a key driver of neurocognitive decline.^42,43 The present findings raise the intriguing possibility that transient and chronic hypercapnia may contribute to such cognitive deficits at least in part through the impairment of perivascular clearance of interstitial solutes.

One final implication of the present findings is the importance of monitoring and maintaining physiological blood gas values in experimental studies of glymphatic exchange and lymphatic drainage, a practice that has not been routinely reported in recent studies defining these physiological processes.^{13,14,16,20,30,44} In this study, we show that without intubation and adequate ventilation, mice develop significant derangements in oxygen saturation, paO₂, and paCO₂ that contribute to marked suppression of solute exchange between the brain and CSF and lymphatic absorption via the cervical lymphatic vasculature. We believe that labs should establish ventilation protocols empirically based on ABG measurements, since dead space volume and ventilator parameters may vary from lab to lab.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by grants from the National Institutes of Health (NIH) to Jeffrey J Iliff (AG054456, NS089709) and to James R Goodman (AG060681). The fluorescence microscopy in this study was further supported by an NINDS P30 grant (NS061800; Aicher PI).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

These experiments were conducted in the Department of Anesthesiology and Perioperative Medicine at Oregon Health & Science University in Portland, Oregon. JRG and JJI conceived the hypotheses, experimental design, and analysis protocols. JRG collected the data. JRG and JJI interpreted the data and drafted the manuscript.

Supplemental material

Supplemental material for this paper can be found at the journal website: http://journals.sagepub.com/home/jcb.