Electrocalcium coupling in brain capillaries: Rapidly traveling electrical signals ignite local calcium signals

In our previous paper, we speculated that electrical (hyperpolarizing) signaling (mediated by Kir2.1 channels) and Ca²⁺ signaling (reflecting IP₃R-mediated release of Ca²⁺ from intracellular stores) intersect in central nervous system (CNS) cECs such that hyperpolarization increases the driving force for Ca²⁺ influx, resulting in an increase in Ca²⁺ signals (5).

To first confirm that activation of Kir2.1 channels causes sufficient hyperpolarization of CNS cECs to produce a relevant increase in the driving force for Ca²⁺ entry, we quantified the changes in membrane potential that accompany activation of Kir2.1 channels using microelectrodes. We focused on cECs within the ACT zone (3/4th-order branch), based on its unique functional role in the CNS microvasculature (2–4, 13). For these experiments, we used an ex vivo, whole-mount, pressurized, retina preparation in which the entire retina tissue is pinned down en face and the ophthalmic artery is cannulated and perfused (8, 14). The advantages of using the retina—an extension of the CNS—is that the resulting ex vivo preparation provides a planar view of the superficial vasculature and allows easy access to cells of interest for impalement (Fig. 1A). Appropriate target cell impalement was confirmed by monitoring the fluorescence of iFluor 488-conjugated hydrazide that diffused from dye-filled microelectrodes into the cEC (Fig. 1A). We found that bath perfusion of the retina preparation with 10 mM K⁺, a concentration within the optimal range for Kir2.1 channel-mediated hyperpolarization (15), and consistent with elevations in extracellular K⁺ that accompany neuronal action potentials (16), hyperpolarized the membrane potential in cECs to −57 ± 4 mV from a resting membrane potential of −29 ± 3 mV. The Goldman–Hodgkin–Katz (GHK) equation predicts that hyperpolarizing the membrane potential to this extent would increase the driving force for Ca²⁺ entry into cECs by 76% (assuming Ca²⁺_o = 2 mM, Ca²⁺_i = 100 nM, and constant permeability), indicating that activation of Kir2.1 channels is capable of strongly increasing Ca²⁺ influx into the cell. The hyperpolarizing effect 10 mM K⁺ was inhibited by bath application of Ba²⁺, a selective blocker of Kir2 channels (17–19) at the concentration used (100 μM) (Fig. 1 B and C).

We previously reported that tetrodotoxin (TTX; 3 μM) and Ba²⁺ (100 μM), applied individually, eliminate ~70 to 80% of baseline Ca²⁺ signals in the barrel cortex, suggesting that these signals are largely dependent on ongoing neuronal activity and Kir channel function (5). To confirm that cEC Kir2.1 channels are specifically responsible for the generation of these Ca²⁺ signals, we developed a tamoxifen-inducible, EC-specific Kir2.1-knockout (KO) mouse (Cdh5-Cre^ERT2; Kir2.1^fl/fl; hereafter, EC Kir2.1-KO mice), crossed with Cdh5-GCaMP8 mice expressing the Ca²⁺ biosensor, GCaMP8. Anesthetized mice carrying the appropriate genetic elements (Materials and Methods) were fitted with a cranial window, and Ca²⁺ signals in cECs were subsequently examined in layer II/III of the barrel cortex using 2-photon laser-scanning fluorescence microscopy. Using a refinement of our original analytical approach (Materials and Methods), we found that ~70% of ongoing (baseline) Ca²⁺ signals in a 425 × 425 μm field of view (FOV) were eliminated in mice lacking Kir2.1 channels in ECs (Fig. 1 D and E and Movie S1). These results, which are virtually identical to those obtained previously using Ba²⁺ and reanalyzed here (SI Appendix, Fig. S1 A and B and Movie S2), confirm a central role for cEC Kir2.1 channels in the neuronal activity-dependent generation of cEC Ca²⁺ signals and establish the basic features of E-Ca coupling.

To directly assess E-Ca coupling in response to a defined, acute neuronal stimulus rather than to ongoing neuronal activity of indeterminate origin, we focally activated cEC Kir2.1 channels by picospritzing 10 mM K⁺ (mimicking neuronal activity) adjacent to an ACT zone capillary segment using a picrospritzer micropipette. We then measured cEC Ca²⁺ responses in the entire FOV at ~2.5 Hz for 236 s. Localized application of 10 mM K⁺ increased cEC Ca²⁺ responses by ~35% within the 425 × 425 μm FOV (Fig. 1 F and G and Movie S3). Taken together with results obtained under basal conditions in EC Kir2.1-KO mice, these findings support the concept that hyperpolarizing signals transmitted by cECs evoke Ca²⁺ signals at various points along the transmission path. This provides a mechanism for increasing the frequency of local Ca²⁺ signals at distances from the original stimulus site that we have dubbed “pseudopropagation,” reflecting its passing resemblance to cell-to-cell conduction of Ca²⁺ signals and potential to exert impacts at a distance from the stimulus site.

To better appreciate the distances and directions over which hyperpolarizing signals spread through capillary networks, we examined arteriole-to-venule distances through capillary beds and capillary branching patterns (Fig. 2 and Movie S4). Distance measurements were performed on 3D models of the cortical vasculature and distances were measured as “vessel distance”—involving tracking through vessels until a particular point was reached. The average 3D distance between capillary branch points was 31.7 µm (Fig. 2 A and B), a value in reasonable agreement with our previous calculations (35.6 µm) and the length of an individual EC (34.9 µm) (see ref. 5). The median distance between arterioles and venules obtained using a 3D flood-fill approach (Materials and Methods) was 263.8 µm, encompassing ~8 ECs (Fig. 2C). These values are comparable to those reported in a previous rigorous brain microvasculature analysis by Kleinfeld et al. which estimated 6 to 8 branches from arteriole to venule (20).

We next examined the directionality of hyperpolarization spread to determine whether there was any inherent bias for hyperpolarizing signals to travel toward arterioles, as might be expected based on the common conceptualization of NVC mechanisms as promoting upstream (i.e., arteriolar) dilation. If this were the case, disproportionally more Ca²⁺ signals should be observed toward the arteriolar side of capillary beds, particularly in the primary branch of the transition zone, where there is only one route available for hyperpolarizing signals to reach an arteriole. Our analysis showed that Ca²⁺ signals were equally distributed on the arteriolar and venule side of capillary beds, with the highest density of Ca²⁺ signals detected at ~200 μm from the nearest major vessel (Fig. 2 D and E)—the midpoint of the capillary bed between arteriolar and venule sides where the number of branches is maximum. This lack of detectable directional bias in cEC Ca²⁺ events toward arterioles or venules implies that Kir2.1-mediated hyperpolarizing signals propagate in all directions following direct focal stimulation of the microvasculature rather than favoring an arteriolar path.

Although the biophysical properties of the Kir2.1 channel make it uniquely suited to sense extracellular K⁺ changes/neuronal activity and hyperpolarize cECs, it is not the only ion channel expressed in these cells capable of exerting a hyperpolarizing influence; K_ATP channels are also present in these cells (8). In this previous study, we reported that activation of K_ATP channels with the synthetic opener, pinacidil (10 μM), robustly hyperpolarized the membrane potential, shifting it from approximately −30 mV to −80 mV. This ~50-mV hyperpolarization would be predicted to increase Ca²⁺ influx by ~150%, based on the GHK equation.

To determine whether K_ATP channel-mediated hyperpolarization, like Kir2.1-mediated hyperpolarization, drives Ca²⁺ signaling in vivo, we first used 2-photon laser-scanning fluorescence microscopy to examine ongoing (baseline) Ca²⁺ signals in the barrel cortex of cranial window model Cdh5-GCaMP8 mice, testing the effects of glibenclamide, a selective synthetic K_ATP channel inhibitor. Glibenclamide (10 μM) significantly attenuated baseline cEC Ca²⁺ events, reducing them by ~50% (Fig. 3 A and B and Movie S5). We have found that cAMP-dependent protein kinase (PKA) is the primary activator of K_ATP channels in vascular cells (e.g., SMCs, pericytes, ECs), reflecting the stimulation of AC and production of cAMP following activation of G_sPCRs by agonists such as adenosine and CGRP (8, 9, 21). Accordingly, we tested the effect of blocking K_ATP channel activation on Ca²⁺ signaling in Cdh5-GCaMP8 mice by inhibiting upstream AC activity with SQ22536 (300 μM). This maneuver dramatically decreased Ca²⁺ signals (by ~80%) (Fig. 3 C and D), suggesting that G_sPCR agonists released as a consequence of local neural activity drive cEC Ca²⁺ signals via K_ATP-dependent hyperpolarization. Collectively, these observations indicate that cEC K_ATP channels exert a strong influence on baseline cEC Ca²⁺ signaling.

We next measured cEC Ca²⁺ signals in vivo in response to focal stimulation of cEC K_ATP channels with pinacidil (10 μM). Picospritzing pinacidil around ACT zone capillary branches caused an increase in cEC Ca²⁺ signaling throughout the FOV that propagated bidirectionally toward arterioles and venules from the site of stimulation (Fig. 3 E and F and Movie S6), similar to Kir2.1-mediated signaling. Focal stimulation of K_ATP channels induced approximately a 50% increase in Ca²⁺ signals in the FOV. Taken together, these results indicate that local activation of K_ATP channels generates Ca²⁺ signals throughout the FOV, suggesting propagating membrane hyperpolarization.

Unlike Kir2.1 channels, K_ATP channels are not activated by hyperpolarization and thus lack the intrinsic capacity to amplify propagating current. This led us to speculate that the ability of K_ATP activation to drive long-distance Ca²⁺ signals reflects its hyperpolarization-induced activation of cEC Kir2.1 channels and engagement of E-Ca coupling. To test this, we focally activated K_ATP channels under conditions in which Kir2.1 channels were genetically eliminated (EC-Kir2.1-KO mice) or blocked (100 μM Ba²⁺) and measured cEC Ca²⁺ responses throughout the FOV. In both EC-Kir2.1-KO mice and in the presence of Ba²⁺, picospritzed pinacidil (10 μM) failed to significantly increase cEC Ca²⁺ signals (Fig. 3 G–I), indicating that K_ATP channel-mediated induction of distant Ca²⁺ signaling is dependent on Kir2.1 channels.

Multiple neural activity-derived compounds are agonists of G_qPCR pathways in cECs. As a testament to the significance of this ongoing production and release of G_qPCR ligands, we previously found that brain-wide inhibition of baseline G_qPCR signaling with the Gα_q/11 inhibitory depsipeptide, FR900359, virtually eliminated (~90% reduction) cEC Ca²⁺ signals in vivo (5). Because this chemical inhibitor might also inhibit Gα_q/11 receptors expressed in other cell types involved in NVC, including neurons, astrocytes, and SMCs (22–25), we sought to confirm that the cEC Ca²⁺ signals observed under baseline conditions were specifically attributable to G_qPCR signaling in cECs. To this end, we used tamoxifen-inducible, brain EC-specific Gα_q/11-KO mice (hereafter, EC Gα_q/11-KO mice), in which Cre expression is driven by the Slco1c1 promoter, which is selectively active in CNS ECs (26). To confirm the functional deletion of Gα_q/11 genes in ECs in our experimental context, we measured changes in somatosensory cortex CBF in response to surface application of the G_qPCR agonist, carbachol (10 μM), using laser-Doppler flowmetry. As expected, carbachol led to a robust increase in CBF in control mice, but not in EC Gα_q/11-KO mice (SI Appendix, Fig. S2 A and B). Consistent with results obtained using the pharmacological inhibitor, FR900359, we found that cEC Ca²⁺ signaling under basal conditions was strongly suppressed (~70%) in EC Gα_q/11-KO mice (Fig. 4 A and B and Movie S7), confirming a crucial role for Gα_q/11 signaling in cECs in the generation of cEC Ca²⁺ signals.

To definitively establish an obligate role for G_qPCR-IP₃/IP₃R signaling in the coupling of hyperpolarization to Ca²⁺ signaling, we measured Ca²⁺ events in vivo after providing a focal hyperpolarizing stimulus by picospritzing 10 mM K⁺ (to activate Kir2.1 channels) or pinacidil (to activate K_ATP channels) onto ACT zone capillary segments in EC Gα_q/11-KO mice. Whereas both of these stimuli produced robust increases in Ca²⁺ signaling in wild-type mice (Figs. 1F and 3E), neither significantly increased cEC Ca²⁺ signals in EC Gα_q/11-KO mice (Fig. 4 C and D). Direct measurements of cEC membrane potential in pressurized retina preparations from EC Gα_q/11-KO mice revealed robust hyperpolarization from baseline following bath application of 10 mM K⁺ (−45 ± 4 to −69 ± 2 mV) or pinacidil (−45 ± 4 to −86 ± 2 mV), indicating that the failure of K⁺ and pinacidil to generate Ca²⁺ signals in EC Gα_q/11-KO mice does not reflect an inability of these agents to effectively hyperpolarize the membrane potential of cECs in these mice (SI Appendix, Fig. S3). Collectively, these findings suggest that on-going G_qPCR signaling in cECs acts as a “priming” step, creating the intracellular preconditions necessary for coupling electrical and Ca²⁺ signaling. They also indicate that hyperpolarization alone—that is, without significant G_qPCR activity, and by extension, IP₃ production (see below)—is insufficient to drive E-Ca coupling.

In our previous studies, we reported that TRPV4 channels are part of an IP₃R-dependent CICR mechanism involved in regulating baseline cEC Ca²⁺ signals (5). Here, we directly tested the involvement of this TRPV4/IP₃R/IP₃R-dependent CICR mechanism in mediating the Ca²⁺-signaling component of E-Ca coupling by measuring Ca²⁺ responses in cECs remote from a focal hyperpolarizing stimulus. To this end, we picospritzed 10 mM K⁺ onto ACT zone capillaries of wild-type Cdh5-GCaMP8 mice, with and without surface application of the selective TRPV4 channel inhibitor, GSK2193874 (hereafter, GSK219; 1 μM), and recorded Ca²⁺ signals. In the absence of GSK219, 10 mM K⁺ and pinacidil induced a significant increase in cEC Ca²⁺ signaling throughout the FOV; this effect was abolished in the presence of GSK219 (Fig. 4 E–H). Collectively, these observations confirm the essential role of TRPV4 channels in mediating the coupling of hyperpolarization with Ca²⁺ signaling. They also indicate that capillary Ca²⁺ events are dependent on G_qPCR-IP₃/IP₃R-TRPV4 channel signaling and reflect the operation of an IP₃R-dependent CICR mechanism.

In our previous report (5), we showed that maneuvers that disrupt intracellular Ca²⁺ signaling, such as depleting ER Ca²⁺ stores by inhibiting SERCA-mediated store refilling with cyclopiazonic acid (CPA), blocking IP₃Rs with xestospongin C or 2-aminoethoxydiphenyl borate, or inhibiting CICR by chelating intracellular Ca²⁺ with EGTA-AM, inhibited ongoing (baseline) Ca²⁺ signals to varying degrees (~55 to 80%), with CPA producing the greatest inhibition (~80%). To determine whether ER Ca²⁺ store release/refilling is similarly central to the E-Ca coupling-dependent Ca²⁺ signaling following a focal hyperpolarizing stimulus, we intravenously injected CPA (0.7 mg/kg) or vehicle into Cdh5-GCaMP8 mice and then measured Ca²⁺ signals throughout the FOV in the barrel cortex after 10 mM K⁺ or pinacidil exposure. Consistent with results obtained for baseline Ca²⁺ signaling, Ca²⁺ signals evoked in response to a hyperpolarizing stimulus were abrogated under conditions in which SERCA activity was inhibited (Fig. 4 I–L), highlighting the importance of intracellular Ca²⁺ store dynamics in the operation of the E-Ca coupling mechanism.

E-Ca coupling can be conceptualized as a rapidly traveling electrical (hyperpolarizing) wave that leaves a trail of “embers”—Ca²⁺ signals—in its wake. This conceptualization invites the application of a genetically engineered voltage-indicator (GEVI) mouse model to directly monitor changes in membrane potential throughout the vasculature in relation to observed Ca²⁺ signals. However, no suitable GEVI mouse model capable of detecting hyperpolarization in ECs is currently available. Thus, we turned to computational modeling to predict the relationship between Kir2.1 and K_ATP stimulation and cEC Ca²⁺ signaling in a simulated microvascular network.

We have previously used an in silico approach to examine how local hyperpolarizing stimuli in the capillary endothelium propagate throughout a reconstructed microvascular network in a cortical volume of 0.5 μL (27). The resulting model comprises 10⁴ cells and simulates membrane potential (V_m) and Ca²⁺ dynamics in cEC, each of which is electrically coupled though gap junctions (Fig. 5A) or saturating pinacidil, allowing currents to propagate along the network. Here, we applied this detailed multicellular model of the cerebral microcirculation to investigate mechanisms that underlie the observed Ca²⁺ activity and to explore the biophysical determinants of E-Ca coupling in capillary networks. Representative simulations (Fig. 5 B and C) performed using a modification of this “virtual voltage-sensor mouse” [see Materials and Methods and (27) for details] predicted stimulatory requirements for physiologically relevant capillary network hyperpolarization. Specifically, an examination of hyperpolarizing activity in a small capillary network segment between a feeding arteriole and a draining venule (Fig. 5B) showed that, under experimental-mimetic conditions (10 mM K⁺ or saturating pinacidil onto 5 to 7 cECs) and using physiologically realistic values for important parameters (i.e., Kir2.1, K_ATP conductances, and gap junctional resistance; see Materials and Methods), focal stimulation was capable of initiating a hyperpolarizing current in the stimulated cells that propagates throughout the capillary network and segments of the penetrating arteriole (PA) and ascending venule (AV).

Computational modeling further corroborated experimental findings that local K_ATP channel-mediated hyperpolarization spreads through the capillary network through subsequent activation of cEC Kir2.1 channels. These simulations showed that focal activation of K_ATP channels with pinacidil (5 to 7 cECs) produces robust, Kir2.1-dependent local and propagating hyperpolarization (Fig. 5B), reaffirming the role of cEC Kir activity as a critical determinant of the ability of the capillary network to propagate hyperpolarizing stimuli. These simulations support a role for the cEC Kir2.1 channel as a mediator of the conducted electrical signal in addition to its role as a K⁺ sensor, and demonstrate how experimentally induced focal stimulation can robustly hyperpolarize the capillary network within the FOV to drive Ca²⁺ activity.

The cEC model (27) was extended to account for stochastic IP₃R dynamics and predict spontaneous Ca²⁺ activity in cECs (Materials and Methods) (Fig. 5A and SI Appendix, Fig. S4A). Fig. 5 C, Top depicts a representative simulation of stochastic Ca²⁺ events from an IP₃R cluster within a cEC. Basal IP₃ levels (reflecting G_qPCR activity) drive stochastic openings of IP₃Rs in the cluster as well as fluctuations in resting cytosolic Ca²⁺ levels, which can be robustly amplified through a CICR mechanism when they cross a threshold value. The resulting pronounced increases in intracellular Ca²⁺ concentration—Ca²⁺ events—are terminated by the Ca²⁺-dependent inactivation of IP₃Rs (SI Appendix, Fig. S4A). Membrane hyperpolarization (Fig. 5 C, Bottom) increases the Ca²⁺ electrochemical gradient and transmembrane Ca²⁺ influx through open NSC channels. The predicted Ca²⁺ influx is physiologically relevant: It can increase basal cytosolic Ca²⁺ levels (Fig. 5 C, Inset) and allows for faster recharging of stores following their depletion during a Ca²⁺ event (Fig. 5C, gray line). The increases in cytosolic Ca²⁺ levels and recharging of Ca²⁺ stores supported by the model increase IP₃R open probability and Ca²⁺ efflux through open IP₃Rs, respectively (Fig. 5C). Taken together, these two effects increase the probability that an IP₃R cluster “fires” and modulates the frequency of Ca²⁺ events in an “active” cEC site. Simulations were performed assuming cECs with varying basal IP₃ concentrations in a simulated FOV representing a slice of brain tissue, 425 × 425 × 50 μm³ volume, containing ~180 cECs (Fig. 5C). The model demonstrates that stochastic opening of IP₃Rs within a cluster in conjunction with CICR results in robust Ca²⁺ events in a subset of “active” cECs with sufficient basal IP₃ levels (Fig. 5D and SI Appendix, Fig. S4B). Following a hyperpolarizing stimulus (10 mM K⁺ onto 5 to 7 cECs), the number of Ca²⁺ events increased by ~280%, from ~17 events/min to 48 events/min (Fig. 5D). Notably, the simulated increase in Ca²⁺ activity with membrane hyperpolarization is in agreement with experiments in which we inhibited or initiated electrical signaling, supporting the conclusion that the E-Ca mechanism can robustly modulate Ca²⁺ activity (Fig. 1 E and G). The model also shows that hyperpolarization-induced transmembrane Ca²⁺ influx translates to more frequent Ca²⁺ events in active sites, but also the recruitment of sites with lower basal IP₃ levels in the simulated FOV (Fig. 5D and SI Appendix, Fig. S4C). The overall increase in the prevalence of Ca²⁺ events within the capillary network can be visualized as a conducted hyperpolarizing wave igniting Ca²⁺ signals in its wake (Movie S8). The model highlights the central role of IP₃ signaling and transmembrane Ca²⁺ influx in modulating Ca²⁺ activity and reinforces experimental results on the role of store recharging. It also provides mechanistic explanations for salient features of this activity at cellular and network levels.

The Kir2.1 channel is uniquely suited to detecting neuronal activity and translating it into a vascular response. No other channel shares its combination of dual activation by extracellular K⁺ and hyperpolarization (17). The specific sensitivity of Kir2.1 channels to activation by extracellular K⁺ enables these channels to respond to K⁺ released into the extracellular space by neurons during an action potential. Because Kir2.1 channels are activated by hyperpolarization and are expressed in all cECs, the hyperpolarization produced by their activation in one cell has the potential to activate these channels in an adjacent cell, setting the stage for cell-to-cell regenerative propagation of the original hyperpolarization signal (1, 27). One important implication of this property is that hyperpolarization produced by any mechanism can activate Kir2.1 channels and “piggyback” on the regenerative potential of these channels, enabling otherwise local signals to propagate.

Although cell-to-cell propagation of Ca²⁺ signals within the vasculature has been reported to occur in certain contexts (28–31), no general mechanism for long-distance transfer of Ca²⁺ signals between vascular cells has been convincingly demonstrated. Evidence has been presented for an extracellular mediator/membrane receptor mechanism (e.g., ATP/Panx1) (29) which is enabled by Kir2.1 channels (32). Our recent work mirrors this theme, showing that neuronal activity-dependent Ca²⁺ signals in brain cECs, monitored in vivo, do not travel from cell to cell but instead are confined to the cell in which they are initiated (5). What does travel from cell to cell is membrane hyperpolarization. And as it does, it superimposes an additional electrical driving force upon the inherent electrochemical gradient for Ca²⁺ in cECs, as described by the GHK relationship. The resulting augmented electrochemical gradient intensifies the driving force for Ca²⁺ entry, increasing Ca²⁺ influx into cECs along the path of the propagating hyperpolarization. In a very real sense, an initial hyperpolarizing signal is converted to Ca²⁺ signals that manifest stochastically at sites along the stimulated microvasculature that can be far removed from the original stimulus.

G_qPCR signaling serves a key priming function by establishing the necessary preconditions that enable an increase in the driving force for Ca²⁺ entry to produce an optically detectable Ca²⁺ signal. The centrality of G_qPCR signaling in this E-Ca coupling response is predicted by model results and corroborated by the finding that long-distance Ca²⁺ signaling was largely absent in EC Gα_q/11-KO mice and virtually eliminated by the selective Gα_q/11 inhibitor, FR900359.

The increase in intracellular IP₃ levels arising from G_qPCR-dependent, PLC-mediated hydrolysis of PIP₂ sensitizes IP₃Rs to intracellular Ca²⁺, providing the molecular impetus for the release of Ca²⁺ from ER stores. Upon binding IP₃, the IP₃R undergoes a conformational change that increases its affinity for Ca²⁺, which binds to a site in a cytoplasm-facing region of the IP₃R. Importantly, Ca²⁺ has dual effects on the IP₃R: At lower concentrations, it serves a CICR function, acting as an agonist to promote release of Ca²⁺ from the ER; at higher concentrations, it acts as an antagonist to inhibit or prevent further release of Ca²⁺ from ER stores (33–35). These dynamics likely enable the initiation and termination of Ca²⁺ mobilization during an event. Model simulations incorporating clusters of IP₃Rs in “G_qPCR-primed” cECs can reproduce the stochastic Ca²⁺ events observed experimentally. Furthermore, model and experimental results demonstrate that hyperpolarization-induced increases in Ca²⁺ influx can increase this Ca²⁺ activity. Intercellular variations in cytosolic/store Ca²⁺ concentration, G_qPCR activity—and thus IP₃—or transmembrane Ca²⁺ permeability can be expected to contribute stochasticity to E-Ca coupling responses. This likely accounts for the observation that Ca²⁺ signals are only detected in a subset of cECs.

Our results suggest that Ca²⁺ influx—the crucial counterpart to sensitized IP₃Rs necessary to complete the CICR machinery—is primarily mediated by the TRPV4 channel. Although the increased electrical driving force produced by membrane hyperpolarization can promote Ca²⁺ entry through any Ca²⁺-permeable ion channel, the number and open probability of these channels need to be appropriate to support sufficient Ca²⁺ entry. In the case of the TRPV4 channel, an important mechanism for increasing open probability is through depletion of PIP₂ by G_qPCR signaling, which disinhibits this channel (7). cECs express multiple types of NSC channels, including TRPV4, TRPA1, Piezo1, and STIM/ORAI1/ORAI3 (5, 29, 36, 37). Among these cation channels, those that are constitutively active can support Ca²⁺ activity through E-Ca coupling, whereas those activated by store or G_qPCR-mediated PIP₂ depletion (ORAI1/ORAI3, TRPV4, TRPA1, TRPC3) may support increase in Ca²⁺ activity independent of electrical signaling.

Our findings also provide a demonstration that K_ATP channel-mediated hyperpolarization intersects with Kir2.1-mediated signaling in cECs to drive CICR-dependent increases in long-range Ca²⁺ signaling. Under physiological conditions, this mechanism would be engaged by neuronally derived agonists, such as adenosine and neuropeptides (e.g., CGRP), which activate G_sPCRs, leading to stimulation of AC, increased cAMP levels/PKA activity, and PKA-mediated activation of K_ATP channels. Collectively, these observations establish a direct link between G_sPCR signaling and the G_qPCR/IP₃/TRPV4 functional module. Interestingly, our data on Ca²⁺ signal induction, hyperpolarization of cECs (retina preparation), and modeling results are in alignment with the idea that focal activation of K_ATP channels in the context of patent Kir2.1 signaling amplifies Kir2.1 channel-mediated hyperpolarization, and thus E-Ca coupling, as reflected in a greater increase in Ca²⁺ signals (50% vs. 35%) and hyperpolarization (−50 mV vs. −28 mV) compared with focal activation of Kir2.1 channels alone. Future studies investigating vascular EC Ca²⁺ signals in mice expressing GEVIs (when they become available), should shed additional light on these relationships.

It has been shown that vascular signals generated by physiological stimuli at the capillary level in the deep parenchyma are capable of retrograde propagation and dilation of upstream arterioles and pial vessels (38–41), findings consistent with our demonstration that focal stimulation (picospritzed K⁺) of capillaries in vivo dilates feeding arterioles via Kir2.1-mediated membrane hyperpolarization (1). The prevailing assumption in the field is that such directional signaling is a hallmark of NVC, and that without it, NVC as we know it would not be possible. Against this backdrop, we found that hyperpolarizing signals generated in vivo by focal stimulation of cECs with K⁺ or pinacidil propagate in all directions from the stimulus site with no bias toward arterioles or venules. While this makes perfect sense given the molecular and electrochemical mechanisms involved and the conceptualization of the cerebral capillary network as a system of wires, it raises the question of how directionality arises in the intact animal. It is possible that excitatory-interneuron circuits, which are bypassed by our focal cEC-stimulation paradigm, impart directionality to the otherwise omnidirectional electrical signal or shape its impact on the vasculature. In addition, contractile pericytes, which are likely electrically coupled to cECs and are strategically positioned upstream of signals arising deeper in the capillary bed, could contribute to directionality by selectively dilating along the retrograde path in response to the passing electrical current.

Interestingly, a branch point/distance analysis following focal stimulation provided insights into the structure of the capillary network, revealing a linear distance between a PA and an AV of ~400 μm, closely corresponding to a single vibrissae cortical volume (~500 μm) (42). This analysis also showed an average of 8 to 10 capillary branches—which is equal to the number of cECs—between an arteriole and a venule, an estimate consistent with a previous seminal analysis of the rodent brain microvasculature by Kleinfeld’s group (20).

Our previous modeling work predicts four possible electrical propagation modes in capillary networks: 1) passive/electrotonic, 2) Kir-facilitated conduction (rapidly developing, like an electric spark), 3) slower and near-simultaneous transitions of a population of cells to an excitable hyperpolarized state (like a slow rising tide), and 4) regenerative signaling transmitted over long distances (like a propagating wave) (27). This last mechanism is akin to the propagation of an action potential. The type of spreading behavior predicted depends primarily on Kir2.1 current density (G_Kir), the number of cECs activated, and gap junctional resistance (R_gj). In a previous study, we modeled hyperpolarization of upstream arterioles based on experiments employing an ex vivo capillary-PA (CaPA) preparation (27) in which 10 mM K⁺ is picospritzed onto the extremities of capillaries attached to an upstream, pressurized arteriole segment—an experimental paradigm that closely parallels the in vivo experimental approach used here. Using G_Kir values estimated from these simulations, and holding other parameter values constant, the model predicts regenerative propagation. In contrast, values of G_Kir estimated from single-cell electrophysiology experiments generally fall in a range that would fail to support this propagation mode. Thus, at this point, it is not entirely clear which propagation mode Kir2.1 promotes and under what conditions. However, evidence suggests that the propagation of hyperpolarizing signals in capillary networks does not simply reflect passive spread (27). Consistent with limited passive spread of charge, Picospritzing a depolarizing concentration of K⁺ (60 mM) onto capillaries in a CaPA preparation does not constrict the upstream arteriole (1).

The ability of Kir2.1 channels to mediate significant amplification of a propagating signal reflects their steep activation by hyperpolarization. Because K_ATP channels are not activated by hyperpolarization, activation of these channels should not lead to regenerative hyperpolarization on its own; thus, K_ATP channel-induced hyperpolarizing signals are limited to passive spread in the absence of Kir2.1 channels. Indeed, we did not detect an increase in Ca²⁺ activity by K_ATP activation in the absence of cEC Kir2.1 channels.

A prominent unanswered—or incompletely answered—question is how G_qPCR activity-dependent PIP₂ depletion, which is known to have reciprocal effects on Kir2.1 channels (inhibition) and TRPV4 channels (disinhibition), fits into the E-Ca coupling framework. Our findings in this regard are somewhat paradoxical. Whereas PIP₂ depletion seems to be intimately involved in enhancing TRPV4 activity and thereby facilitating Ca²⁺ influx, CICR, and store refilling, the predicted inhibition of Kir2 activity was not evident in our experimental paradigm. Differential kinetics of Kir2.1 inhibition and TRPV4 disinhibition is one possible explanation. Our previous single-cell electrophysiology studies showed that inhibition of Kir2.1 channel activity by PIP₂ depletion takes up to 20 min to fully manifest (6), a time frame that would place this inhibitory effect outside our experimental window (~6 min). Literature reports that this inhibition occurs within seconds in other cell types (e.g., neurons) (43, 44) and expression systems add intrigue to this question.

The role of pericytes in E-Ca coupling is not known. Our previous work (2) and earlier studies by Puro and colleagues (45, 46) suggest that pericytes and endothelial cells are electrically coupled. However, the nature of coupling and its bidirectionally are unclear. It is possible that pericytes could act as current sinks for electrical signaling in cECs, hindering electrical signaling in cECs by passing depolarizing currents into them. They could also act as amplifiers of cEC electrical signaling based on the nature and magnitude of pericyte K+ currents (47). These are active areas of investigation.

How E-Ca coupling impacts physiological processes is an area of active investigation. The increased driving force for Ca²⁺ entry into cECs that occurs as a consequence of propagating hyperpolarization from sites of neural activity in response to physiological stimuli increases the spatiotemporal prevalence of Ca²⁺-release sites throughout the capillary network. This could aid in providing sufficient Ca²⁺ for local blood flow control (e.g., via an eNOS/NO mechanism) along the often-considerable distances from the sites of neural activity to the feeding arteriole and AV. Hemodynamic responses to single-whisker stimulation propagate beyond the principle cortical volume (42, 48–50), potentially perfusing the entire column and beyond. E-Ca coupling, which provides an elegant mechanism for coordinating Ca²⁺ events by linking hyperpolarization and enhanced Ca²⁺ signaling, could ensure the effective and directed routing of blood through the tangle of capillaries to areas in need. Taking the finding of omnidirectional electrical signaling at face value, the downstream spread of this signal could also contribute to increasing venous EC permeability, possibly aiding in the clearance of metabolic waste or supporting immune cell transcytosis. In addition to these roles, E-Ca coupling may also be involved in resolving capillary stalls. In summary, our study recasts capillary endothelial cells as excitable cells and the coupling between excitation (hyperpolarization) and Ca²⁺ bridges the gap between fast, long-range electrical signaling and slower, local Ca²⁺ events that supply and direct blood flow to areas of need.

Statistical testing was performed using GraphPad Prism 10 software. Data are expressed as means ± SEM, and a P-value ≤ 0.05 was considered significant. All experiments were paired, unless otherwise stated. Data were tested for normality using a Shapiro–Wilk test prior to application of the appropriate parametric or nonparametric statistical test. Stars denote significant differences; “ns” indicates nonsignificant comparisons. Statistical tests are noted in figure legends. All t tests were two-sided. Sample sizes were estimated based on similar experiments performed previously in our laboratory. Data collection was not performed blinded to the conditions of the experiments. Littermates were randomly assigned to experimental groups; no further randomization was performed.

A.M., N.M.T., D.H.-E., and M.T.N. designed research; A.M., T.H., and N.M.T. performed research; A.M., G.W.H., T.H., and N.M.T. analyzed data; and A.M., G.W.H., N.M.T., D.H.-E., and M.T.N. wrote the paper.

The authors declare no competing interest.