Brain metabolic state dictates the polarity of astrocyte control over the cerebrovasculature

Abstract

Calcium signaling in astrocytes couples changes in brain activity to regional alterations in cerebral blood flow (CBF) by eliciting vasoconstriction or vasodilation of adjacent arterioles^1-7. However, the mechanism for how these disparate astrocyte influences provide appropriate changes in cerebral vascular tone within a cellular environment that has dynamic metabolic requirements remains unclear. The regulation of CBF has recently been shown to be tightly coupled to the lactate/pyruvate ratio and thus the NADH/NAD⁺ ratio in animals⁸ and humans^9,10. We tested the impact of metabolic changes on the regulation of cerebral blood vessel diameter by astrocytes, by manipulating tissue oxygenation which changes dynamically with brain activity^11-14. Using two-photon Ca²⁺ imaging and uncaging as well as intrinsic nicotinamide adenine dinucleotide (NADH) imaging of single cells as a measure of redox state, we show that the ability of astrocytes to induce vasodilations over vasoconstrictions critically relies on the metabolic state of the tissue. When O₂ availability is lowered and astrocyte [Ca²⁺]_i is elevated, astrocyte glycolysis and lactate release are maximized. Extracellular lactate contributes to prostaglandin E₂ (PGE₂) accumulation by hindering its transporter-mediated uptake, subsequently causing vasodilation. These data reveal the role of metabolic substrates in regulating CBF and provide a mechanism for differential astrocyte control over cerebrovascular diameter during different states of brain activation.

We first examined whether changing PO₂ consistently determined whether vasoconstrictions or vasodilations were evoked by astrocyte Ca²⁺ transients. The mGluR agonist t-ACPD (100 μM) potently increased [Ca²⁺]_i (240.0 ± 19.1 %, n = 14, P < 0.01, Fig. 1 a) in astrocytes^1,4,5 and differential responses were observed in the adjacent cerebral blood vessels whereby vasodilation (107.0 ± 0.8 % (100% = control diameter), P < 0.01, n = 23, Fig. 1 a, d) occurred in low O₂ but in high O₂ arterioles constricted¹ (85.4 ± 4.0 %, P < 0.01, n = 10, Fig. 1 d). We next used two-photon photolysis of the Ca²⁺ cage DMNPE-4¹⁵ to examine the effect of low O₂ levels on the impact of directly evoked astrocyte Ca²⁺ signals on adjacent blood vessels. Uncaging DMNPE-4 triggered Ca²⁺ waves that spread to multiple endfeet circumscribing arterioles (Fig. 1 b, c). In contrast to previous results in high PO₂ where Ca²⁺ transients in astrocyte endfeet always caused constriction¹, in low O₂ endfeet Ca²⁺ (232.2 ± 8.5 %, n = 17, P < 0.01) caused arteriole dilations (107.4 ± 1.0 %, P < 0.01, n = 17, Fig. 1 b, c, d) that were repeatable (Fig. 1 b). The Ca²⁺ change preceded the enlargement of lumen diameter (Fig. 1 b, c). These data suggest PO₂ dictates the polarity of the change in arteriole diameter caused by either mGluR activity or an elevation in endfoot Ca²⁺ and that the dilation process can be initiated solely by astrocyte activation (see Supplemental Fig. 1 for PO₂ measurements). Ca²⁺ activates cPLA₂ triggering arachidonic acid (AA) formation that is either converted to 20-HETE¹ in (SMCs) causing vasoconstriction or is converted to the vasodilator PGE₂ in astrocytes via cyclooxygenase (COX) enzymes^4,7. We confirmed COX expression in astrocytes and their endfeet (Supplemental Fig. 2). The COX inhibitor indomethacin (100 μM) blocked vasodilations caused by t-ACPD (98.4 ± 0.8 %, P > 0.05, n = 9, Fig. 1 d) and by caged Ca²⁺ photolysis (101.0 ± 0.5 %, P > 0.05, n = 6, Fig. 1 d) in low O₂. Application of the COX product PGE₂ (1 μM) elicited vasodilation (109.4 ± 2.7 %, P < 0.05, n = 4, Fig. 1 d), confirming COX activation and the generation of PGE₂ is an important signaling molecule in astrocyte-mediated vasodilations in low O₂^4,7.

Fig. 1.

Lowering O₂ converts astrocyte-mediated vasoconstrictions to vasodilations. (a) Top: astrocytes (red) loaded with Ca²⁺ indicator surround arteriole. Bottom: astrocyte Ca²⁺ signals occur coincident with dilation caused by the mGluR agonist t-ACPD in low O₂. (b) Top: uncaging astrocyte Ca²⁺ (indicated by arrow) causes vasodilation in low O₂ and is repeatable. Bottom: expanded time scale shows the Ca²⁺ signal in endfoot 1 (shown in c I) precedes the lumen diameter increase. (c) Overlay of vessel and pseudo colored endfoot Ca²⁺ changes. Images I-V correspond to times in b. I: control state and regions of interest: endfeet 1-3. Image II: endfoot 1 shows first Ca²⁺ rise (star) before lumen diameter starts to increase. Vertical dotted line (III-V) indicates previous position of vessel wall. (d) Summary data.

Next we examined how changing tissue PO₂ could modify whether astrocyte Ca²⁺ transients induce constrictions versus dilations. We examined the possibility that a lower PO₂ elevates anaerobic metabolism, increasing external lactate which subsequently enhances external PGE₂ concentrations. Once produced, PGE₂ passively diffuses into the extracellular space¹⁶ and its action on smooth muscle cells is terminated when prostaglandin transporters (PGTs) take up PGE₂ by an exchange of intracellular lactate¹⁷. Due to the influence of a lactate concentration gradient on PGT efficacy, we tested the hypothesis that higher levels of extracellular lactate reduce the ability of PGT to uptake PGE₂¹⁷ from the extracellular space, thereby increasing PGE₂ concentrations resulting in vasodilation. Such a role for lactate is consistent with the observations that extracellular lactate levels increase immediately during brain activity¹⁸ and that lactate is known to enhance activity dependent increases in blood flow⁸. We found using immuohistochemistry that PGT is widely expressed in brain grey matter in both astrocyte endfeet (Fig. 2 a), and neurons (Fig. 2 b), indicating dendrites as well as endfeet take up prostaglandins. Consistent with a positive correlation between external lactate and PGE₂, we found that low O₂ enhanced lactate release from brain slices (low: 114.2 ± 9.1 μM, n = 6; high: 41.9 ± 5.6 μM, n = 6, P < 0.01, Fig. 2 c). Also, significantly greater levels of PGE₂ were observed in low O₂ when PGE₂ production was triggered by mGluR activation, (low: 136.1 ± 10.2 pg/ml, n = 6; high: 91.9 ± 12.6 pg/ml, n = 4, P < 0.01, Fig. 2 d). Furthermore, addition of lactate (1 mM) enhanced the PGE₂ level (control: 40.5 ± 3.3 pg/ml; + lactate: 58.0 ± 3.1 pg/ml, n = 5, P < 0.01) and increased arteriole diameter (107.5 ± 1.0 %, P < 0.01, n = 15, Fig. 2 e, f, g). Dilations induced by lactate were blocked by indomethacin (100.4 ± 0.4 %, P > 0.05, n = 11, Fig. 2 g) indicating they were mediated by PGE₂. These data indicate that in low O₂ the higher level of extracellular lactate raises external levels of PGE₂. This is consistent with lactate reducing PGE₂ uptake through PGT.

Fig. 2.

Low O₂ facilitates lactate and PGE₂ release and enhances astrocyte glycolysis. (a and b) Immunohistochemistry showing the astrocyte marker GFAP and the neuron marker MAP2 colocalize with PGT labeling. (c) Lactate release is elevated in low O₂. (d) t-APCD increases PGE₂ release most in low O₂. (e and f) Lactate dilates an arteriole. Images I-III correspond to time points in f, which shows the lumen diameter increase. (g) Summary showing percent dilation by lactate and block by indomethacin. (h) Colocalization of the astrocyte maker SR-101 (left stack) and the NADH signal (middle stack) from perivascular glia somas and endfeet. (i) Top: arteriole NADH fluorescence from a single image plane showing astrocyte compartments, SMCs/endothelial cells and empty lumen of the blood vessel. Middle and Lower: pseudo colored image of NADH fluorescence in high and low O₂. (j) Summary of NADH changes in astrocyte compartments caused by low O₂. (k and l) Astrocyte NADH in response to glycolysis inhibition with iodoacetate (IA) (k) and LDH inhibition with Oxamate (l). Left: single experiment; right: summary.

We next examined the role of glycolysis in lactate production by imaging the intrinsic fluorescence of the metabolic electron carrier NADH in astrocytes. Two-photon excitation of NADH provides a sensitive, subcellular measure of both oxidative metabolism (punctate NADH fluorescence from mitochondria) and glycolytic metabolism (diffuse NADH fluorescence from the cytosol)¹⁹. We observed, as previously reported, that astrocytes (stained with SR-101²⁰) showed bright intracellularly diffuse NADH fluorescence in the soma and endfeet (Fig. 2 h, i). Stimulating glycolysis with low O₂ increased the astrocyte NADH signal (124.6 ± 1.4 %, n = 7, P < 0.01, Fig. 2 i, j) and inhibiting glycolysis with iodoacetate (IA, 200 μM) reduced basal NADH fluorescence (87.8 ± 2.0 %, n = 5, P < 0.05, Fig. 2 k). Furthermore, inhibiting lactate dehydrogenase (LDH) (which converts pyruvate and NADH to lactate and NAD+; see Fig. 4 e diagram) with oxamate (2.5 mM) increased the NADH signal (119.3 ± 2.0 %, n = 6, P < 0.01, Fig. 2 l). These data support the conclusion that NADH is generated to a significant extent in astrocytes as a consequence of glycolytic metabolism—likely originating from glycogen breakdown²¹—which can be augmented by reducing O₂.

Fig. 4.

Raising PGE₂ levels by inhibiting PGT changes the polarity of astrocyte-mediating vasomotion. (a) PGE₂ levels were further elevated by mGluR activation in low O₂ when PGT was inhibited by U46619 or T34. (b) Summary data of vasomotion during PGT manipulation. (c) Uncaging Ca²⁺ in high O₂ causes vasodilation in exogenous lactate. Top panels show the astrocyte Ca²⁺ signal change (pseudo colour) from uncaging. Lower panels show close up of vessel lumen. (d) In high O₂, astrocyte-mediated vasoconstriction is converted to vasodilation during PGT blockade. Left: astrocytes and endfeet circumscribing an arteriole. Ca²⁺ is uncaged in 3 astrocytes separated in time; box indicates the vessel region examined on the right. Right: vasomotions corresponding to the separate uncaging events. Small pseudo colour images show the Ca²⁺ signal change from uncaging in each astrocyte. Lower images of the vessel and endfeet (red) show that the vasomotion switches polarity when PGTs are blocked. (e) Diagram of the supported model.

Recent two-photon NADH imaging demonstrated an increase in astrocyte glycolysis subsequent to the onset of neuronal activity¹⁹. We hypothesized that mGluR activation further enhances glycolysis in astrocytes in low O₂, promoting greater extracellular lactate levels that cause vasodilation. Consistent with this idea, t-ACPD enhanced extracellular lactate, which was greatest in low O₂ (low: 186.7 ± 11.2 μM, n = 6; high: 98.6 ± 10.2 μM, n = 7, P < 0.01, Fig. 3 f). The astrocytes were a site of increased glycolysis because t-ACPD triggered an increase in cytosolic NADH (128.7 ± 4.1 %, n = 7, P < 0.01, Fig. 3 a, b, c, d, e) coincident with lumen widening (108.5 ± 0.7 %, n = 5 of 7, P < 0.01, Fig. 3 b, c). In contrast, peak NADH signals from perivascular neurons were not as great and were qualitatively different, showing distinct punctate mitochondrial fluorescence (Supplemental Fig. 3). These data signify a change in anaerobic metabolism may be important for astrocyte-mediated vasomotion in low O₂.

Fig. 3.

Glycolysis and lactate release is critical for astrocyte-mediated dilations. (a) Arteriole and astrocyte NADH image in low O₂; bottom shows ROIs for astrocyte compartments (1 and 2) and vessel (box). (b) Pseudo color NADH (top) and diameter (bottom) changes caused by t-ACPD at time points I-III in c. Vertical dotted line (II and III) indicates previous position of vessel wall. (c) Astrocyte NADH (top) and lumen diameter (bottom) in response to t-ACPD; same experiment as a and b. (d) Summary of NADH increase from mGluR activation in low O₂. (e) Top: NADH image of an arteriole and astrocyte. Lower: soma close-up showing t-ACPD causes a diffuse increase in NADH. (f) t-ACPD increases extracellular lactate most in low O₂. (g) t-ACPD decreases astrocyte NADH during glycolysis inhibition; top: single experiment; bottom: summary. (h) mGluR activation increases astrocyte NADH during LDH inhibition; top: single experiment; bottom: summary. (i) Summary showing t-ACPD fails to dilate vessels in oxamate or IA and PGE₂ rescues vasodilation in these compounds. (j and k) The increase in lactate (j) and PGE₂ (k) caused by t-ACPD is significantly less in oxamate and IA.

Recent functional imaging data from human subjects reveals a relationship between a rise in the lactate/pyruvate and thus NADH/NAD⁺ ratios and an increase in CBF⁹, but how this enhanced glycolytic state is a critical step translating information on the level of brain activity into vasodilation of arterioles is not known. Our data demonstrate that lactate, astrocyte NADH and PGE₂ are maximally increased in low O₂ when mGluRs are activated and that exogenous lactate can mediate prostaglandin dependent vasodilation. Therefore, we tested the hypothesis that glycolysis and thus lactate production is essential for prostaglandin dependent astrocyte-mediated vasodilations. We used two separate pharmacological treatments to limit the lactate released by mGluR activity in low O₂. First, in the presence of IA to block the source of lactate, t-ACPD instead decreased astrocyte NADH (81 ± 0.4 %, n = 5, P < 0.01, Fig. 3 g) and failed to dilate vessels (97.8 ± 0.2 %, n = 5, P > 0.05, Fig. 3 i). Second, in the presence of oxamate to curtail lactate formation, t-ACPD still increased astrocyte NADH (113.8 ± 3.1 %, n = 6, P < 0.05, Fig. 3 h)—an expected outcome as oxamate acts downstream of NADH production (see Fig. 4 c diagram)—yet vasodilations no longer occurred (99.0 ± 0.1 %, n = 6, P > 0.05, Fig. 3 i). We then confirmed the lack of vasomotion observed during these treatments was associated with a significant reduction in lactate and PGE₂ release when mGluRs were activated (Lactate release in t-ACPD: 180.2 ± 11.9 μM, n = 6; + IA: 88.2 ± 8.8 μM, n = 6, P < 0.01; + Oxamate: 89.1 ± 6.6 μM, n = 6, P < 0.01, Fig. 3 j)(PGE₂ release in t-ACPD: 154.0 ± 10.1 pg/ml, n = 4; + IA: 115.4 ± 10.6 pg/ml, n = 6, P < 0.01; + Oxamate: 94.6 ± 7.6 pg/ml, n = 6, P < 0.01, Fig. 3 k). Finally, because PGE₂ is the final effector molecule on SMCs (downstream of astrocyte glycolysis), we could rescue vasodilation by applying PGE₂ in the presence of either oxamate (109.2 ± 2.2 %, P < 005, n = 5, Fig. 3 i) or IA (109.0 ± 1.3 %, P < 0.01, n = 5, Fig. 3 i). These data demonstrate that inhibition of glycolysis or LDH limits extracellular lactate and PGE₂ accumulation in response to t-ACPD in low O₂, preventing astrocyte mediated vasodilations.

As a final test of our hypothesis we pharmacologically targeted PGTs to manipulate extracellular levels of PGE₂ and examined the effects on vasomotion. First, blockade of PGTs with U46619²² or TGBz T34¹⁶ resulted in elevated levels of PGE₂ when mGluRs were activated in low O₂ (tACPD: 227.5 ± 7.2 %, tACPD + U46619: 435.6 ± 16.9, P < 0.01 to tACPD alone, n = 5; tACPD + T34: 816.4 ± 6.0, P < 0.01 to tACPD alone, n = 4, Fig. 4 a) and did not interfere with low O₂ astrocyte-mediated dilations (U46619: 110.4 ± 0.3 %, P < 0.01, n = 5; T34: 109.1 ± 1.3 %, P < 0.01, n = 7, Fig. 4 b). Second, we hypothesized that adding exogenous lactate in high O₂ would enable astrocyte-mediated vasomotion to change polarity by attenuating PGE₂ uptake. However, because there is nothing preventing the initiation of vasoconstriction in high O₂ from the action of AA in SMCs, we blocked constriction at the level of SMCs by adding adenosine²³. Extracellular adenosine is elevated in a reduced oxygen environment²⁴ and is an additional modulator of vasoconstriction²³. In the presence of adenosine (100 μM), uncaging Ca²⁺ within astrocytes in high O₂ produced no vasomotion (99.6 ± 0.9 %, P > 0.05, n = 4, Supplemental Fig. 4). However, when lactate (100 μM - 1 mM) was included to mitigate PGE₂ clearance in addition to adenosine, Ca²⁺ uncaging now resulted in dilation (107.5 ± 1.4 %, P < 0.01, n = 9, Fig. 4 b, c), despite equivalent endfeet Ca²⁺ signals compared to adenosine alone (adenosine: F/F₀ = 160.4 ± 6.1; adenosine + lactate: F/F₀ = 159.6 ± 5.1, P > 0.05, Supplemental Fig. 4). From this, we hypothesized that PGT blockade would have a similar result by raising the extracellular PGE₂ level. Indeed, in high O₂ we found that in the same vessel, the first uncaging event caused constriction (87.4 ± 1.8 %, P < 0.01, Fig. 4 b, d) but the second uncaging event during PGT inhibition with T34 caused dilation (108.7 ± 1.4 %, P < 0.05, n = 5, Fig. 4 b, d). These data indicate that the ability of astrocytes to induce dilation over constriction critically relies on reduced efficacy of PGE₂ uptake under conditions where lactate release is enhanced.

Here we show that when the glycolytic state of the brain is enhanced by lower O₂ levels, astrocyte-mediated vasoconstrictions convert to vasodilations because the elevated extracellular lactate causes PGE₂ accumulation which is maximized during mGluR activation. Flow through blood vessels is proportional to vessel radius to the 4^th power; therefore a ~9 % change in arteriole diameter reported here equates to a ~45 % increase in CBF. This change is consistent with astrocyte-mediated vasodilations observed in vivo⁷ and can more than account for regional CBF changes measured by PET²⁵ and with two-photon microscopy²⁶ when the brain is physiologically activated. These data add an important and until now unrecognized mechanism in the bidirectional control of cerebral vessels by astrocytes. Previous studies implicated different populations of GABA neurons²⁷ and tissue nitric oxide levels². The metabolic control of these antagonistic pathways by astrocytes implies that brain blood flow would reach a homeostatic level depending on metabolic activity. Under quiescent periods when O₂ is not being rapidly consumed, astrocyte Ca²⁺ signals induce constrictor tone, keeping CBF at an appropriate level. At the onset of neural activity there is a drop in PO₂ and rise in extracellular lactate which promotes astrocyte mediated dilation. Manipulating this homeostatic balance may be a therapeutic avenue for treating the inappropriate declines in CBF that occurs in some dementias and after stroke.

Methods Summary

Brain Slices and Imaging

Hippocampal and neocortical slices were prepared from juvenile (p 16-21), male, Sprague-Dawley rats. Care and use of animals was approved by the University of British Columbia Animal Care and Use Committee. ACSF bubbled with 95 % O₂, 5 % CO₂ (standard for brain slice experimentation), was defined as a high O₂ treatment and ACSF bubbled with 20 % O₂, 5 % CO₂, balanced N₂, was defined as a low O₂ treatment. For astrocyte dye loading, slices were transferred to a 3 ml ACSF well containing the Ca²⁺ cage DMNPE-4/AM and/or the Ca²⁺ indicator rhod-2/AM dissolved in DMSO (final DMSO concentration 0.2 %) for 1.5 hours. A two-photon laser-scanning microscope (Zeiss LSM510-Axioskop-2 fitted with a 40X-W/0.80 numerical aperture objective lens) directly coupled to a Chameleon Ultra tunable ultrafast laser (~100-fs pulses and 76MHz, Coherent) provided excitation of rhod-2 intrinsic NADH fluorescence and was used for uncaging using two photon photolysis. Arterioles were identified with IR-DIC optics and vessel diameter changes were imaged by acquiring the transmitted laser light. Quantification of lumen diameter, NADH and Ca²⁺ changes were performed with Zeiss LSM (version 3.2) software and ImageJ.

Lactate and PGE₂ Measurements

The assay used for measuring PGE₂ release was Specific Parameter PGE₂ ELISA kits (R&D systems). Extracellular lactate levels were measured using a Lactate Assay Kit (Biomedical Research Service Centre, SUNY Buffalo).

Full Methods and associated references can be found below in the supplemental materials section.