Distinct signatures of calcium activity in brain mural cells
Abstract
Pericytes have been implicated in various neuropathologies, yet little is known about their function and signaling pathways in health. Here, we characterized calcium dynamics of cortical mural cells in anesthetized or awake Pdgfrb-CreERT2;Rosa26< LSL-GCaMP6s > mice and in acute brain slices. Smooth muscle cells (SMCs) and ensheathing pericytes (EPs), also named as terminal vascular SMCs, revealed similar calcium dynamics in vivo. In contrast, calcium signals in capillary pericytes (CPs) were irregular, higher in frequency, and occurred in cellular microdomains. In the absence of the vessel constricting agent U46619 in acute slices, SMCs and EPs revealed only sparse calcium signals, whereas CPs retained their spontaneous calcium activity. Interestingly, chemogenetic activation of neurons in vivo and acute elevations of extracellular potassium in brain slices strongly decreased calcium activity in CPs. We propose that neuronal activation and an extracellular increase in potassium suppress calcium activity in CPs, likely mediated by Kir2.2 and KATP channels.
Introduction
The entire abluminal surface of the cerebral vasculature is covered by mural cells, namely vascular smooth muscle cells (SMC) and pericytes, which exhibit a continuum of phenotypes. Despite several attempts to categorize these heterogeneous cells in the adult mouse by morphology and molecular footprints (Hill et al., 2015; Hartmann et al., 2015a; Hartmann et al., 2015b; Vanlandewijck et al., 2018), their identities remain unclear. In general, SMCs are ring-shaped cells, which express α-smooth muscle actin (αSMA) and surround arteries and penetrating arterioles. Pericytes are embedded within the vascular basement membrane and are classically described to have a protruding ‘bump on a log’ cell body with processes that run longitudinally along capillaries (Attwell et al., 2016). Many morphologically different subtypes of pericytes have been described (Grant et al., 2019; Uemura et al., 2020).
Pericytes regulate vascular morphogenesis and maturation as well as maintenance of the blood–brain barrier (Armulik et al., 2011; Armulik et al., 2010; Daneman et al., 2010). In the context of vascular development, several signaling pathways between endothelial cells (EC) and pericytes are important; for instance, in the recruitment of mural cells through platelet-derived growth factor B (PDGFB) and platelet-derived growth factor receptor beta (PDGFRβ) signaling (Gaengel et al., 2009), as well as Angiopoietin/Tie2 signaling in angiogenesis (Teichert et al., 2017). Furthermore, CNS pericytes in the mature cortex may provide a basal tone to the vasculature and contribute to vessel stability (Berthiaume et al., 2018). In recent years, pathologies such as diabetic retinopathy, Alzheimer’s disease, circulatory failure, and primary familial brain calcification have been attributed to the loss or dysfunction of pericytes (Kisler et al., 2017; Liu et al., 2019; Montagne et al., 2018; Winkler et al., 2014; Zarb et al., 2019; Nikolakopoulou et al., 2019). Yet, little is known about the function and signaling properties of pericytes in the healthy brain.
Being part of the neurovascular unit (NVU), SMCs and pericytes are in close contact with astrocytes, neurons, oligodendrocytes, and microglia. The NVU is involved in intricate regulatory mechanisms to tightly control blood flow (Iadecola, 2017). One of these mechanisms is functional hyperemia, which couples neural activity with changes in cerebral blood flow (CBF). Also, intrinsic vascular tone oscillations, known as vasomotion, have been observed maintaining blood ‘flowmotion’ in the brain at rest (Intaglietta, 2017). While it is evident that there is a relation between vasomotion and cytosolic calcium levels in SMCs (Filosa et al., 2006; Longden et al., 2016), participation of pericytes in vasomotion is currently debated (Hill et al., 2015; Hall et al., 2014; Fernández-Klett et al., 2010; Peppiatt et al., 2006). A recent study suggests that pericytes located at junctions of postarteriole transition regions are able to regulate blood flow through the capillary network (Gonzales et al., 2020). Moreover, specific optogenetic stimulation of capillary pericytes (CPs) could demonstrate a slow capillary constriction, suggesting that CPs might contribute to basal blood flow resistance (Hartmann et al., 2021).
Here, we combined in vivo and ex vivo approaches to investigate calcium signaling of mural cells in the somatosensory cortex vasculature of healthy adult mice. Given the distinct locations of morphologically diverse mural cells on the vasculature (Hartmann et al., 2015a; Grant et al., 2019), we wondered whether these cells would differ in their calcium signaling properties. Furthermore, we investigated how different stimuli such as vasomodulators and neuronal activity impact the calcium dynamics of CPs.
Results
Two-photon imaging of Pdgfrb-driven GCaMP6s in mural cells
To study calcium dynamics in mural cells, we crossed Pdgfrb-CreERT2 mice (Gerl et al., 2015) with Rosa26< LSL-GCaMP6s> (Ai96) reporter mice (Figure 1A). Measurements were performed in anesthetized and awake mice, as well as acute brain slices. For localization and classification of mural cells, we defined the vessel types based on their branch order and diameter. The continuum of mural cells along the arteriovenous axis was categorized, as described earlier (Grant et al., 2019), into SMC (0th branch), ensheathing pericytes (EP) (1st–4th branch), CP (>4th branch), and venular pericytes (VP) at postcapillary venules (Figure 1B, Figure 1—figure supplement 1A-F). The term EPs is used synonymous with precapillary or terminal vascular SMCs (Hill et al., 2015; Hartmann et al., 2021). Noteworthy is that in rare cases CPs interconnected vessel segments by extending processes through the parenchyma (Figure 1—figure supplement 1D). These few interconnecting pericytes were excluded from our analysis, since at the time it could not be determined, whether they are remnants of vessel regression or whether they represent a structural feature with importance in neurovascular coupling (Brown, 2010; Alarcon-Martinez et al., 2020; Corliss et al., 2020). Besides mural cells, some GCaMP6s-expressing astrocytes were sparsely detected (Figure 1B) and verified by labeling with astrocyte dye sulforhodamine 101 (SR101) (Figure 1—figure supplement 2). To avoid overlap of pericytic calcium signals by astrocytic signals, we omitted regions where a differentiation between GCaMP6s signals from astrocytes and pericytes was not possible.

Two-photon calcium imaging of mural cells.
(A) GCaMP6s expression in Pdgfrb-positive cells of Pdgfrb-CreERT2:R26-GCaMP6sf/stop/f transgenic mice, induced by four consecutive Tamoxifen injections. If required, adeno-associated viruses (AAV) were injected before the chronic cranial window implantation over the somatosensory cortex. Images were acquired at a wavelength of 940 nm, and the vasculature was labeled via an intravenous injection (iv.) of 2.5% Texas Red Dextran (70 kDa). In vivo experiments were conducted with awake and anesthetized (1.2% isoflurane, supplied by a ventilation mask) mice. For pharmacologic interventions, acute brain slices of the same mice were prepared. Z-stacks of the vessel arbor were acquired to determine the precise location of the imaged cells along the vasculature. (B) In vivo images of the cortical vasculature showing GCaMP6s (green) labeled mural cells. SMCs are located on pial arteries and ensheathing pericytes on 1st–4th branch order vessels. CPs (consisting of mesh and thin stranded pericytes) are found on vessels of >4th branch order. Occasionally, some astrocytes showed GCaMP6s expression. VPs reside at postcapillary venules. Scale bars: 10 μm. See also Figure 1—figure supplements 1 and 2.
Distinct basal calcium transients of mural cells in vivo
In line with previous reports (Hartmann et al., 2015a; Grant et al., 2019), we found SMCs on surface arterioles and penetrating arterioles with an average inner diameter of 18.0 ± 7.3 μm (values are mean ± SD) (Figure 2—figure supplement 1). EPs were located on vessels of 1st–4th branch order with an average inner diameter of 7.3 ± 1.7 μm. CPs were found on capillaries (>4th branch order) with an average diameter of 4.1 ± 0.7 μm (Figure 2—figure supplement 1).
In lightly anesthetized animals (1.2% isoflurane), all the mural cells described above revealed basal calcium fluctuations in somata and processes (Figure 2A–D, Video 1, Video 2, Video 3, Video 4). Initial visual inspections of the calcium traces from SMCs and EPs indicated similar calcium dynamics as both revealed synchronous calcium signals (Figure 2A and B). Contrarily, CPs and VPs exhibited asynchronous calcium signals that are also appearing in microdomains along the processes (Figure 2C and D).

Mural cell calcium dynamics in vivo.
(A–D) Representative images of (A) SMCs, (B) EP, (C) CP, and (D) VP in vivo. Regions of interest (ROIs) for somata (S, magenta) were hand selected. ROIs for processes (P1–3, blue) were found in an unbiased way, employing a MATLAB-based algorithm (see Figure 2—figure supplement 3). Below the images are the normalized time traces of calcium signals of GCaMP6s fluorescence extracted from the corresponding example ROIs. Vessel diameters were measured with line scans across somata (indicated by the dashed yellow lines). Scale bars: 10 µm. (E–G) Plots depicting the relation between cytosolic calcium (green trace) and vessel diameter (red trace) measured simultaneously with line scans along the soma, for (E) SMC, (F) EP, and (G) CP. Traces were centered on the average diameter of the vessel. (H) Comparison of the correlations (Fisher z-transformed) between calcium signals and changes in vessel diameter. Unpaired two-tailed t-tests were performed to compare groups. Individual values with mean ± SD are shown. SMC: n = 13, EP: n = 13, CP: n = 9. t(24) = 1.139, p=0.27; t(20) = 14.21, ***p<0.001. (I–K) Quantification and comparison of baseline calcium signal properties: (I) frequency (signals per minute), (J) peak amplitude (df/f), and (K) duration (s) between SMC, EP, and CP somata and processes of in vivoanest mice, shown as violin/box plots (Tukey whiskers). The dashed yellow lines indicate the mean values of the respective parameters. Statistics were calculated using linear mixed-effects models and Tukey post hoc tests, SMCanest: N = 14, n = 67, EPanest: N = 16, n = 53, CPanest: N = 25, n = 93. *p<0.05, **p<0.01, ***p<0.001. (L–N) Quantification and comparison of baseline calcium signal properties: (L) frequency (signals per minute), (M) peak amplitude (df/f), and (N) duration (s) between SMC, EP, and CP somata and processes of in vivoawake mice, shown as violin/box plots (Tukey whiskers). The dashed yellow lines indicate the mean values of the respective parameters. Statistics were calculated using linear mixed-effects models and Tukey post hoc tests, SMCawake: N = 4, n = 53; EPawake: N = 3, n = 32; CPawake: N = 3, n = 96. *p<0.05, **p<0.01, ***p<0.001. n.s. indicates not significant. See also Figure 2—figure supplements 1–4. Data: Table 1 and Figure 2—source data 1.
-
Figure 2—source data 1
Source data for Figure 2.
- https://cdn.elifesciences.org/articles/70591/elife-70591-fig2-data1-v3.xlsx
Basal SMC calcium activity in vivo.
Recorded at 11.84 Hz and played at 45 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 5 µm.
Basal ensheathing pericyte calcium activity in vivo.
Recorded at 11.84 Hz and played at 45 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 10 µm.
Basal capillary pericyte calcium activity in vivo.
Recorded at 11.84 Hz and played at 45 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 10 µm.
Basal venular pericyte calcium activity in vivo.
Recorded at 11.84 Hz and played at 45 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 10 µm.
Calcium signals in SMCs during spontaneous vasomotion were inversely correlated with vessel diameter (transformed rFisher z = 1.20 ± 0.37, n = 13, Figure 2E and H), as reported in a previous study (Hill et al., 2015). A similar behaviour of calcium signaling was observed in EPs (transformed rFisher z = 1.06 ± 0.22, n = 13, Figure 2F and H). Moreover, power spectral analysis of the EP calcium signals showed a distinct peak at 0.1 Hz (Figure 2—figure supplement 2), which is in line with the vasomotion frequency observed in SMCs (van Veluw et al., 2020; Mateo et al., 2017). In accordance with earlier studies (Pabelick et al., 2001), we found that the oscillations in vessel diameter follow the calcium transients in EPs by an average delay of 300 ms, suggesting a calcium-dependent contraction mechanism. In contrast, there was no relationship between spontaneous calcium signals in CPs and capillary diameter (transformed rFisher z = –0.005 ± 0.06, n = 9, Figure 2G and H).
To compare the calcium dynamics between different mural cells, we measured frequency, amplitude, and duration of calcium transients using semi-automated image analysis as described and employed earlier by our lab (Zuend et al., 2020; Stobart et al., 2018b). Regions of interest (ROI) for pericyte somata were selected by hand and ROIs for pericyte processes were automatically detected with an unbiased algorithm (Figure 2—figure supplement 3; Ellefsen et al., 2014) implemented in a custom-written MATLAB toolbox, CHIPS (Barrett et al., 2018).
In anesthetized animals, calcium signals of SMCs and EP somata were similar in amplitude and duration (Figure 2J and K); however, signal frequency (expressed as mean ± SD signals/min) was higher in EPs compared to SMCs (8.3 ± 2.8 vs. 7.1 ± 1.6, Figure 2I, Table 1). On the other hand, differences in calcium dynamics were more pronounced between EPs and CPs. Compared to EPs, calcium signals in CP processes were two times more frequent and about 30% shorter in duration (Figure 2I and K), whereas calcium signals in CP somata were almost twofold larger in amplitude and ~20% shorter in duration (Figure 2J and K, Table 1).
Basal calcium signal properties of mural cells in awake and anesthetized animals.
Summary table of calcium signal frequency, amplitude, and duration of different mural cells from awake and anesthetized measurements in vivo. Values are represented as mean ± SD.
Frequency (signals/min) | Amplitude (df/f) | Duration (s) | ||||
---|---|---|---|---|---|---|
Awake | Anesthetized | Awake | Anesthetized | Awake | Anesthetized | |
SMC | 9.4 ± 3.0 | 7.1 ± 1.6 | 0.3 ± 0.2 | 0.6 ± 0.3 | 3.1 ± 1.3 | 3.4 ± 0.9 |
EP_Soma | 8.9 ± 3.2 | 8.3 ± 2.8 | 0.5 ± 0.4 | 0.6 ± 0.4 | 2.6 ± 0.9 | 3.0 ± 0.9 |
EP_Process | 20.2 ± 7.8 | 21.6 ± 12.0 | 0.9 ± 0.5 | 1.3 ± 0.7 | 2.9 ± 0.7 | 3.4 ± 0.9 |
CP_Soma | 6.9 ± 2.7 | 8.5 ± 2.3 | 0.3 ± 0.1 | 1.1 ± 1.3 | 2.4 ± 0.7 | 2.4 ± 0.6 |
CP_Process | 39.6 ± 14.5 | 45.7 ± 19.3 | 0.7 ± 0.3 | 1.3 ± 0.4 | 2.6 ± 0.6 | 2.4 ± 0.4 |
Prolonged isoflurane anesthesia has been shown to affect the cerebral vasculature and cause a decrease in vasomotor activity of arteries and arterioles (van Veluw et al., 2020; Slupe and Kirsch, 2018). We therefore only included measurements of SMCs and EPs up to 25 minutes post-anesthesia induction for analysis. Longer anesthesia (>30 min) led to a stark reduction in calcium activity (data not shown). However, to avoid possible anesthesia-related alterations on basal calcium activity of mural cells, we also performed calcium imaging in awake mice, as previously employed in our lab (Zuend et al., 2020; Stobart et al., 2018a). In the following we refer to anesthetized measurements as in vivoanest and awake measurements as in vivoawake.
Indeed, there were differences in mural cell calcium dynamics between measurements in awake and anesthetized mice (Figure 2—figure supplement 4, Table 1). SMC calcium signals were more frequent in awake mice compared to anesthetized mice, while the calcium signal frequency in EPs and CPs was not significantly affected (Figure 2—figure supplement 4A, D, G, Table 1). Calcium signal amplitudes (expressed as df/f ± SD) in CPs were two to three times lower in awake compared to anesthetized animals, in both somata (in vivoawake: 0.3 ± 0.1 vs. in vivoanest.: 1.1 ± 1.3) and processes (in vivoawake: 0.7 ± 0.3 vs. in vivoanest.: 1.4 ± 0.4, Figure 2—figure supplement 4H, Table 1).
Importantly, the overall differences in calcium signaling signatures between mural cells that were observed in anesthetized mice (Figure 2I–K) were similar to those measured in awake mice (Figure 2L–N), emphasizing that more distal pericytes are functionally distinct. For example, signal frequency was two times higher in CP processes compared to EP processes (39.6 ± 14.5 vs. 20.2 ± 7.8, Figure 2L, Table 1). Also, signal amplitudes and durations differed significantly between CPs and EPs in both their somata and processes (Figure 2M and N, Table 1). Thus, EPs and CPs have distinct basal calcium dynamics, which, besides their location in the vascular network, morphology, and αSMA expression, can be used as a further measure to differentiate between these mural cell subtypes. Furthermore, despite transcriptional similarities between EPs and SMCs, subdomain (soma/process) calcium activity in EPs is a characteristic, which is not present in SMCs.
Persisting calcium signals in CPs ex vivo
To further investigate calcium signal properties of mural cells, we prepared acute brain slices for ex vivo pharmacological probing. Blood plasma was stained with an iv. injection of 70 kDa Texas Red Dextran (100 μL, 2.5%) via the tail vein prior to the slice preparation. The fluorophore remained in the vasculature for several hours for easy capillary identification and further served as a control for motion artifacts. Orientation in the slice was obtained by following a penetrating artery from the slice surface to its branches (Figure 1A).
Interestingly, CPs in acute brain slices retained their highly frequent calcium transients, which were detected in both their somata and processes (Figure 3A). And CP spontaneous calcium signal frequency in both somata and processes in slices was not significantly different to in vivoawake and in vivoanest. basal calcium activity (Figure 3B). However, signal amplitudes in CPs were on average larger in slices compared to in vivoawake and in vivoanest (Figure 3—figure supplement 1).

Mural cell calcium dynamics in acute cortical brain slices.
(A) Representative image of a CP ex vivo. In the GCaMP6s channel, ROIs for soma (S, in magenta) and processes (P1–3, in blue) are shown. On the right are the respective normalized traces of calcium signals. (B) Violin/box plots depicting the quantified calcium signal frequency of CPs ex vivo compared to the previously (Figure 2) determined calcium signal frequency in vivo for somata and processes. The dashed yellow lines indicate the mean values. Statistics were calculated using linear mixed-effects models and Tukey post hoc tests, CPawake: N = 3, n = 96; CPanest: N = 25, n = 93; CPex vivo: N = 52, n = 106, n.s. indicates not significant. Scale bars: 10 μm. (C) SMCs and EP ex vivo can be localized similarly to those in vivo by following the vessel branches of a pial artery lying on the surface of the brain slice. Z-stack (20 μm) image of an arteriole and its adjacent 1st-order branch, where SMCs and an EP is located. The yellow box shows a magnified image of the EP and the cyan box shows a magnified image of SMCs. In the GCaMP6s channel, ROIs for soma (S, in magenta) and processes (P1–3, in blue) are shown. (D) Corresponding normalized traces of calcium signals for the EP and SMCs in (C) are shown. (E) Z-stack (20 μm) images of an arteriole and its adjacent 1st-order branch harboring SMCs and an EP (indicated by the yellow circles), before and after U46619 treatment. Next to the images are the corresponding normalized traces of calcium signals for the indicated SMC and EP. (F) Quantifications of the calcium signal frequency in SMC and EP somata and processes, comparing baseline to the U46619 treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using two-tailed paired t-tests, SMC: N = 4, n = 14; EP: N = 4, n = 11, ***p<0.001. Scale bars: 10 μm. (G) Time-averaged (10 s) images of a CP before, during U46619 (100 nM) treatment and recovery (30 min washout of U46619). U46619 causes a massive calcium response, which is accompanied by cytoplasmic blebbing and membrane ruffling (yellow arrowhead). The dashed cyan lines in the TxRed images outline the vessel at baseline level to highlight vessel changes during treatment and recovery. (H) A normalized trace of the calcium response of a CP to U46619 (100 nM) is shown. On the right is the quantification of the area under the curve (auc) comparing baseline to U46619 treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a two-tailed Wilcoxon matched-pairs signed rank test. N = 5, n = 16, ***p<0.001. Scale bars: 5 μm. The red line below the calcium trace indicates the time of drug addition. See also Figure 3—figure supplements 1–3. Data: Figure 3—source data 1.
-
Figure 3—source data 1
Source data for Figure 3.
- https://cdn.elifesciences.org/articles/70591/elife-70591-fig3-data1-v3.xlsx
Surprisingly, in contrast to CPs, calcium activity of SMCs and EPs was greatly reduced ex vivo (Figure 3C and D). Compared to in vivoanest, the calcium signal frequency of EPs was reduced more than fourfold in somata (ex vivo: 1.9 ± 1.7 vs. in vivoanest.: 8.3 ± 2.8) and more than 10-fold in processes (ex vivo: 1.9 ± 2.2 vs. in vivoanest: 21.6 ± 12.0). Similarly, in SMCs, the calcium signal frequency was reduced more than fourfold ex vivo compared to in vivoanest (1.6 ± 1.3 vs. 7.1 ± 1.6). We wondered whether the pronounced reduction of calcium activity in SMCs and EPs is linked to a loss of a vascular tone in slices. Most studies on mural cells in acute slice preparations include the addition of a thromboxane A2 receptor agonist (U46619) to generate an artificial vascular tone (Mishra et al., 2014; Brown et al., 2002). Strikingly, upon bath-application of U46619 (100 nM), calcium signal frequency in SMCs and EPs (both in somata and processes) increased more than fourfold (Figure 3E). Incidentally, a similar effect of U46619 on spontaneous activity in SMCs and proximal pericytes was recently reported in retinal preparations (Gonzales et al., 2020).
Curiously, in CPs, we observed that short pulses (1 min) of U46619 (100 nM) elicited a strong calcium response (Figure 3G and H and Video 5). This overt calcium response in CPs was accompanied by morphological changes, such as membrane ruffling and cytoplasmic blebbings or extrusions (Figure 3, Figure 3—figure supplement 2). These cytoplasmic changes in CPs recovered after U46619 washout (>30 min), suggesting that the transient U46619-mediated activation of CPs was not toxic. Nonetheless, to further investigate calcium signaling properties of CPs, we omitted the use of U46619 in our slice preparations. Furthermore, diameter measurements in brain slices were not performed because two-photon microscopy of stained stationary blood plasma may not reliably detect vessel borders.
Ex vivo application of U46619 (100 nM) leads to a strong calcium response in CPs with cytoplasmic extrusions.
Recorded at 0.75 Hz and played at 20 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 10 µm.
Worthy of note is that in the absence of U46619 VPs also retained their calcium dynamics ex vivo with no evident change in both somata and processes (Figure 3—figure supplement 3), further emphasizing the viability of mural cells in our slice preparations and that mural cells differ in their calcium signaling properties.
CP calcium events are evoked by vasomodulators
Next, we probed CPs for their responsiveness to different vasomodulators, such as G-protein-coupled receptor (GPCR) agonists Endothelin-1 (100 nM), UDP-Glucose (100 μM), and ATP (100 μM). Indeed, all agonists triggered a cytosolic calcium increase in CPs (Figure 4A–C), indicating that CPs express functional GPCRs and can respond to factors that are known to be released by endothelial cells or astrocytes (Dehouck et al., 1997; Lazarowski and Harden, 2015; Harden et al., 2010; Koizumi et al., 2005). Since the tested agonists activate GPCRs, mainly acting via phospholipase C and subsequent IP3-mediated calcium release from internal stores (Wynne et al., 2009; Lazarowski and Harden, 2015; Abbracchio et al., 2006), we continued to examine the involvement of ion channels, which have been shown to affect cytosolic calcium levels in SMCs (Hill-Eubanks et al., 2011). Application of Nimodipine (100 μM), a blocker of L-type voltage-gated calcium channels (VGCC) involved in SMC contractions, led to a moderate reduction of basal calcium activity in CP somata (9.1 ± 2.6 vs. 7.6 ± 1.7, Figure 4D), but not in processes (39.1 ± 8.6 vs. 39.6 ± 7.8, Figure 4D). However, a transient receptor potential channel (TRPC) blocker (Earley and Brayden, 2015), SKF96365 (100 μM), reduced the calcium transient frequency more than threefold in both somata and processes (Figure 4E), suggesting that CPs require extracellular calcium to maintain their basal calcium fluctuations. Control drug vehicle (DMSO) experiments revealed no changes in CP calcium signal frequency in both somata and processes (Figure 4—figure supplement 1). Moreover, to rule out a signal run-down due to cell death when treated with SKF96365, cells were afterwards subjected to a brief stimulation with U46619, which triggered a robust calcium response in CPs (Figure 4E).

Modulation of CP calcium signals ex vivo.
(A–C) CP calcium responses to the application of vasomodulators: (A) Endothelin-1 (100 nM), (B) ATP (100 µM), and (C) UDP-Glucose (100 µM).
On top are representative normalized calcium signal traces and below are quantifications of the area under the curve (auc), comparing baseline to the respective treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using Wilcoxon matched-pairs signed rank tests. Endothelin-1: N = 3, n = 7, *p=0.02; ATP: N = 4, n = 8, **p=0.008; UDP-Glucose: N = 4, n = 11, ***p<0.001. (D) CP calcium response to application of the L-type voltage-gated calcium channel (L-type VGCC) blocker Nimodipine (100 µM). On top is a representative normalized calcium signal trace and below is the quantification of the signal frequency in somata and processes, comparing baseline to the Nimodipine treatment. Nimodipine was infused 15 min prior to data collection. Data represents individual cells, median and interquartile ranges. Statistics were calculated using two-tailed paired t-tests, N = 4, n = 11. S: t(10) = 2.39, *p=0.04; P: t(10) = 0.1495, p=0.88. (E) CP calcium response to application of the TRPC channel blocker SKF96365 (100 µM) and endstimulus application (U46619, 100 nM). On top is a representative normalized calcium signal trace and below is the quantification of the signal frequency in somata and processes, comparing baseline to the SKF96365 treatment and the quantification of the area under the curve (auc), comparing baseline to endstimulus treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a Wilcoxon matched-pairs signed rank test, N = 3, n = 7. S: *p=0.02; P: *p=0.02, endstim: *p=0.02. The red lines below the traces indicate the addition time of the respective drug. n.s. indicates not significant, *p<0.05, **p<0.01, ***p<0.001. See also Figure 4—figure supplement 1, 2 and 3. Data: Figure 4—source data 1.
-
Figure 4—source data 1
Source data for Figure 4.
- https://cdn.elifesciences.org/articles/70591/elife-70591-fig4-data1-v3.xlsx
Neuronal stimulation leads to a calcium signal drop in mural cells from capillaries to upstream arterioles
To investigate how acute changes in neuronal activity impact mural cell calcium dynamics, we used a chemogenetic approach (Roth, 2016), in which transduced neurons expressing hM3D(Gq)-DREADD (designer receptors exclusively activated by designer drugs) were activated with 30 µg/kg iv. clozapine (Jendryka et al., 2019). Clozapine was chosen over clozapine N-oxide (CNO) due to better bioavailability and the absence of side effects in the here applied concentration (Gomez et al., 2017; Cho et al., 2020). We injected 50 nl of AAV2-hSyn-hm3D(Gq)-mCherry into the somatosensory cortex with the aim to limit the transduced area as such, that along a continuous vascular branch the area of hM3D(Gq)-DREADD expressing neurons was primarily within the capillary bed (Figure 5A and B). This allowed us to investigate how neuronal activation alters mural cell calcium responses along a connected vascular branch from capillaries to arterioles and arteries.

Calcium signal drop in all mural cells along a vascular branch in response to neuronal activation.
(A) Z-stack (10 µm) image of a CP inside the area of neurons, expressing hM3D(Gq)-mCherry. (B) Z-stack (75 µm) of a vascular tree with connected capillaries reaching into an area of neurons expressing hM3D(Gq)-mCherry (indicated by the dashed cyan lines), before and during clozapine treatment. Scale bar = 10 µm. (C) Normalized calcium signal trace of a CP to chemogenetic activation of hM3D(Gq) transduced neurons in vivo and the quantification of the calcium signal frequency, comparing baseline to stimulation. Data represents individual cells, median, and interquartile ranges. Statistics were calculated using a Wilcoxon matched-pairs signed rank test for S (***p<0.001) and a two-tailed paired t-test for P (t(15) = 8.815, ***p<0.001), N = 4, n = 16. (D) Images of a CP before and during clozapine treatment. The numbers indicate the vascular diameter in µm. On the right vascular diameters measured at the site of the soma, before (4.1 ± 0.8 µm) and during treatment (4.5 ± 0.7 µm) are shown. Data represents individual cells, mean and standard deviation. Statistics were calculated using a two-tailed paired t-test, t(15) = 3.245, **p=0.005, N = 4, n = 16. Scale bar = 5 µm. (E) Normalized calcium signal trace of an EP (located on a vascular branch connected to capillaries in an area of hM3D(Gq) transduced neurons) to chemogenetic activation of neurons in vivo. On the left quantification of the calcium signal frequency, comparing baseline to stimulation is shown. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a two-tailed paired t-test for S (t(10) = 10.71, ***p<0.001) and a Wilcoxon matched-pairs signed rank test for P (***p<0.001), N = 3, n = 11. (F) Images of two EPs (somata are indicated by yellow arrowheads) before and during clozapine treatment are shown. The numbers indicate the vascular diameter in µm. On the right vascular diameters measured at the site of the soma, before (5.9 ± 1.1 µm) and during treatment (7.5 ± 1.9 µm) are shown. Data represents individual cells, mean and standard deviation. Statistics were calculated using a two-tailed paired t-test, t(10) = 5.375, ***p<0.001, N = 3, n = 11. Scale bar = 5 µm. (G) Normalized calcium signal trace of a SMC (located on a vascular branch connected to capillaries in an area of hM3D(Gq) transduced neurons) to chemogenetic activation of neurons in vivo. On the left quantification of the calcium signal frequency, comparing baseline to stimulation is shown. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a two-tailed paired t-test, t(12) = 16.55, ***p<0.001, N = 3, n = 13. (H) Images of SMCs before and during clozapine treatment are shown. The numbers indicate the vascular diameter in µm. On the right vascular diameter before (14.7 ± 6.6 µm) and during treatment (18.2 ± 8.1 µm) are shown. Data represents individual cells, mean and standard deviation. Statistics were calculated using a two-tailed paired t-test, t(12) = 6.985, ***p<0.001, N = 3, n = 13. Scale bar = 20 µm. See also Figure 5—figure supplements 1 and 2. Data: Figure 5—source data 1.
-
Figure 5—source data 1
Source data for Figure 5.
- https://cdn.elifesciences.org/articles/70591/elife-70591-fig5-data1-v3.xlsx
Chemogenetic activation of neurons completely diminished calcium activity in CPs in both somata and processes (Figure 5C, Video 6). Possible non-specific effects caused by clozapine on CP calcium signals were ruled out because CPs outside the area of DREADD-expressing neurons retained their spontaneous calcium dynamics (Figure 5—figure supplement 1). A similar calcium drop in pericytes in response to neuronal activity was recently reported in the olfactory bulb upon odorant stimulation (Rungta et al., 2018). Interestingly, we observed that the drop in CP calcium signals was accompanied by a capillary diameter increase of 8% compared to baseline (Figure 5D). We further monitored calcium changes in upstream EPs and SMCs in the arterioles and pial artery connected to the investigated capillaries. These upstream segments were outside of the DREADD expressing area, however, although we could not detect mCherry signals we cannot exclude projections of neurons expressing low levels of hM3D(Gq) in these upstream segments. Nonetheless, in response to chemogenetic activation of neurons predominantly located in the vicinity of the capillary bed, also EPs exhibited a clear drop in calcium signals in both somata and processes (Figure 5E). And the diameter of EP-associated arterioles increased by 22% compared to baseline (Figure 5F). Furthermore, in the upstream connected pial artery, SMCs also exhibited a calcium signal drop (Figure 5G) and arterial diameter increased by about 19% (Figure 5H). Importantly, to rule out that the calcium signal drop is not caused by toxicity associated with clozapine induced DREADD activation, we monitored the same cells one day after the clozapine treatment and found a complete recovery of calcium activity in all of the observed mural cells (Figure 5—figure supplement 2). Thus, chemogenetic neuronal activation led to a pronounced calcium drop in all mural cells along the same vascular tree and to an increase in vessel diameters, from the capillary to the surface pial artery.
Chemogenetic activation of hM3D(Gq) DREADD transduced neurons leads to a calcium signal drop in CPs.
Arrows point to CPs. Recorded at 11.84 Hz and played at 45 fps. Scale bar 10 µm.
Elevations in extracellular potassium lead to a calcium signal drop in CPs
When stimulated, neurons release potassium into the extracellular space (Somjen, 1979; Heinemann and Lux, 1977). This potassium is sensed by SMCs, causing the suppression of calcium oscillations (Filosa et al., 2006). To test whether calcium signals also drop in CPs to elevations in extracellular potassium, we temporarily raised the extracellular potassium concentration in acute brain slices by 10 mM (in the form of KCl). Indeed, CPs revealed a pronounced drop in calcium activity in both somata and processes in response to potassium stimulation (Figure 6A). To exclude any secondary stimulation effects, neuronal firing activity was blocked with 1 µM tetrodotoxin (TTX) (Figure 6—figure supplement 1). TTX alone did not influence the calcium dynamics of CPs in both somata and processes (Figure 6—figure supplement 2).

CPs react to increased extracellular potassium.
(A) Calcium response of CPs to a 10 mM rise of extracellular potassium in the presence of TTX (1 µM) ex vivo. A representative normalized calcium signal trace (left) and the quantification of the calcium signal frequency, comparing baseline to treatment (right) are shown. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a Wilcoxon matched-pairs signed rank test (for S) and a two-tailed paired t-test (for P), N = 4, n = 13. S: ***p<0.001; P: t(12) = 8.516, ***p<0.001. (B) Calcium response of CPs to a 10 mM rise of extracellular potassium in the presence of TTX (1 µM) and Kir2 channel blocker BaCl2 (100 µM) ex vivo. A representative normalized calcium signal trace is shown on the left and on the right, the calcium signal frequency is quantified, comparing baseline to pre-treatment (TTX + BaCl2) and additional potassium treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a one-way ANOVA, N = 4, n = 10. S: F(1.796, 16.17) = 5.728, p=0.02, P(baseline vs. BaCl2) = 0.82, p(baseline vs. BaCl2+ KCl) = 0.10, p(BaCl2 vs. BaCl2+ KCl) = 0.009; P: F(1.542, 13.88) = 7.137, p=0.01, p(baseline vs. BaCl2) = 0.31, p(baseline vs. BaCl2+ KCl) = 0.22, P(BaCl2 vs. BaCl2+ KCl) = 0.002. (C) Calcium response of CPs to a 10 mM rise of extracellular potassium in the presence of TTX (1 µM), Kir2 channel blocker BaCl2 (100 µM), and KATP channel blocker PNU-37883A (100 µM) ex vivo. A representative normalized calcium signal trace is shown on the left. On the right, the calcium signal frequency is quantified, comparing baseline to pre-treatment (TTX + BaCl2+ PNU-37883A) and additional potassium treatment. Data represents individual cells, median and interquartile ranges. Statistics were calculated using a one-way ANOVA, N = 4, n = 12. S: F(1.395, 15.35) = 0.2248, p=0.72, P(baseline vs. PNU + BaCl2) = 0.76, p(baseline vs. PNU + BaCl2+ KCl) = 0.99, p(PNU + BaCl2 vs. PNU + BaCl2+ KCl) = 0.77; P: F(1.683, 18.51) = 0.9687, p=0.38, p(baseline vs. PNU + BaCl2) = 0.96, P(baseline vs. PNU + BaCl2+ KCl) = 0.54, P(PNU + BaCl2 vs. PNU + BaCl2+ KCl) = 0.44. (D) Calcium response of CPs to NaN3 (5 mM) ex vivo. A representative normalized calcium signal trace is shown on the left and the quantification of the calcium signal frequency, comparing baseline to NaN3 treatment is shown on the right. Data represents individual cells, median and interquartile ranges. Statistics were calculated using two-tailed paired t-tests, N = 3, n = 7. S: t(6) = 4.824, **p=0.003; P: t(6) = 7.621, ***p<0.001. (E) Calcium response of CPs to an acute hypoxia insult in vivo. On the left, two-photon images show the time course and GCaMP6s fluorescence of the hypoxic insult (yellow arrowhead: capillary pericyte; white arrowhead: astrocyte). Center, a representative normalized calcium signal trace is shown. The calcium signal frequency is quantified on the right. Baseline is compared to hypoxic intervention. Data represents individual cells, median and interquartile ranges. Statistics were calculated using Wilcoxon matched-pairs signed rank tests, N = 4, n = 16. S: ***p<0.001; P: ***p<0.001. The red lines indicate the addition time of the respective drug. See also Figure 6—figure supplements 1–3. Data: Figure 6—source data 1.
-
Figure 6—source data 1
Source data for Figure 6.
- https://cdn.elifesciences.org/articles/70591/elife-70591-fig6-data1-v3.xlsx
In SMCs and ECs, Kir2 channels are the main drivers for potassium sensing (Longden and Nelson, 2015; Longden et al., 2017). Single-cell transcriptomic analysis revealed a high abundance of Kir2.2 (Kcnj12) transcripts in pericytes (Vanlandewijck et al., 2018). Hence, it could be that a similar mode of potassium sensing occurs also in pericytes. We antagonized Kir2 channels with BaCl2 (100 µM), and interestingly, we observed a less pronounced decrease in CP calcium frequency in response to potassium stimulation in both somata (7.7 ± 2.5 vs. 4.6 ± 2.7) and processes (55.1 ± 27.9 vs. 43.2 ± 33.2, Figure 6B).
Furthermore, an ATP-sensitive potassium channel (Nichols, 2006) composed of Kir6.1 (Kcnj8) and its associated sulfonylurea receptor 2 (SUR2, Abcc9) has been suggested as a marker for brain pericytes due to its high abundance in CPs (Bondjers et al., 2006; Vanlandewijck et al., 2018) and was recently shown to be involved in CP hyperpolarization (Sancho et al., 2021). Application of PNU-37883A (100 µM), a Kir6.1 specific antagonist (Teramoto, 2006), together with the Kir2 channel blocker BaCl2 abolished the decrease in CP calcium signals in response to potassium stimulation, in both somata (9.1 ± 3.2 vs. 9.0 ± 4.0) and processes (46.9 ± 31.1 vs. 55.0 ± 36.4, Figure 6C).
Next, we investigated the possible involvement of KATP channels in the regulation of calcium signaling in CPs. Lowering the cellular [ATP]/[ADP] ratio by blocking mitochondrial activity with sodium azide (NaN3, 5 mM) in ex vivo brain slices should result in KATP channel activation (Dart and Standen, 1995; Tinker et al., 2018; Reiner et al., 1990; Lerchundi et al., 2019). Indeed, CP calcium signals decreased strongly in response to NaN3 in both somata and processes (Figure 6D). To test whether reducing mitochondrial respiration impacts CP calcium dynamics also in vivo, we exposed anesthetized animals to an acute hypoxia insult, by replacing oxygen in the anesthesia gas mixture transiently with pure nitrogen. Indeed, transient hypoxia led to a complete cessation of CP calcium activity in both somata and processes (Figure 6E). When oxygen supply was re-established, calcium activity in CPs was resumed. This hypoxia-induced drop in calcium activity was specific to mural cells, since nearby GCaMP6s-labeled astrocytes reacted with a strong increase in calcium levels upon hypoxia (Figure 6E). Additionally, we assessed whether capillary diameters changed during the acute hypoxic challenge. Indeed, during hypoxia capillary diameters dilated on average by 11% compared to baseline (Figure 6—figure supplement 3). When re-establishing the oxygen supply, capillary diameters returned to baseline values. Overall, these capillary diameter changes coincided with the CP calcium signal changes during and after hypoxia (Figure 6E, Figure 6—figure supplement 3).
Discussion
The present study identifies distinct calcium signatures of different mural cells in the somatosensory cortex of healthy mice. To differentiate between these phenotypically diverse mural cells, we employed a previously coined classification based on microvascular localization: SMC, EP, and CP (Grant et al., 2019; Berthiaume et al., 2021). Although phenotypic diversity of mural cells was already outlined in the earliest descriptions of pericytes almost a century ago (Zimmermann, 1923), there is still no consensus on mural cell classifications in the brain (Attwell et al., 2016). Especially mural cells located at the terminal end of arterioles (1st–4th branch order) are not clearly defined and were variously named in recent literature as precapillary/terminal SMCs (Hill et al., 2015), contractile mural cells (Hariharan et al., 2020), contractile pericytes (Gonzales et al., 2020), or EPs (Berthiaume et al., 2021). Whether these cells are a subtype of SMCs or pericytes remains to be elucidated.
CPs have been reported to have frequent calcium transients in somata and in microdomains along processes (Hill et al., 2015; Rungta et al., 2018). We have extended these observations and found that all mural cells along the arteriovenous axis exhibit basal calcium fluctuations (Figure 2A–D, Video 1, Video 2, Video 3, Video 4). During spontaneous vasomotor activity, SMCs exhibited calcium oscillations that are inversely correlated to vessel diameter changes (Figure 2E). These rhythmic pulsations were shown to be based on entrainment by ultra-slow fluctuations (0.1 Hz) in γ-band neuronal activity (Mateo et al., 2017). We observed the same behavior of calcium oscillations in EPs on penetrating arterioles (1st–4th branch order, Figure 2F). This is in line with the expression of α-smooth muscle actin in mural cells of up to 4th branch order vessels as seen by immunohistochemistry staining and the use of transgenic animals (Hartmann et al., 2015a; Hill et al., 2015). In recent computational and experimental works, these vascular segments have also been shown to be the location of blood flow regulation to supply a particular downstream region (Sweeney et al., 2018; Grubb et al., 2019; Gonzales et al., 2020; Video 7). Thus, it is likely that conclusions in implicating pericytes in the control of blood flow in several studies were based on experiments focusing on EPs (Peppiatt et al., 2006; Hall et al., 2014). At the capillary level, which we defined as vessels with an average diameter of 4 μm and a branch order of >4, CPs exhibit asynchronous calcium signals between somata and processes (Figure 2C). In agreement with previous studies, we did not find an immediate correlation between vessel diameter and calcium transients in CPs in vivo under basal conditions (Figure 2G; Hill et al., 2015; Rungta et al., 2018). Nonetheless, we have to note that two-photon imaging is not suitable for the detection of minute and fast changes in vessel diameter in the expected range of 1–3% for capillaries (Grutzendler and Nedergaard, 2019).
Mural cells at a precapillary branch.
Recorded at 11.84 Hz and played at 45 fps, showing Merge, TxRed, and GCaMP6s channels. Scale bar 10 µm.
Interestingly, CP calcium signal frequency persisted in ex vivo slice experiments with no substantial difference from the signal frequency in vivo (Figure 3A and B). This could hint to a blood flow independent signal, which is rather influenced by the content of solutes and oxygen, available in excess to the cells in the slice preparation. In contrast, SMCs and EPs showed a pronounced reduction in their calcium signal frequency ex vivo (Figure 3C and D). Due to the lack of intraluminal pressure in acute brain slices, arteries and arterioles lose basal tone, resulting in vascular collapse (Mishra et al., 2014). This likely led to cessation of basal calcium oscillations in SMCs and EPs ex vivo (Figure 3C and D). Addition of a widely used preconstricting factor (U46619, 100 nM) to the brain slice (Brown et al., 2002; Filosa et al., 2006) restored calcium activity in both SMCs and EPs (Figure 3E and F). However, we found that addition of U46619 caused CPs to form membranous blebs and cellular extrusions as well as provoking a large cytosolic calcium increase (Figure 3G and H, Video 5). This aberrant calcium response was accompanied by vessel constriction, which were also reported earlier (Fernández-Klett et al., 2010). It is likely that CPs are able to exert force on their underlying vessel in pathological conditions via cytoskeletal rearrangements. A comparable scenario was suggested in Aβ-induced constriction of capillaries in Alzheimer’s disease (Nortley et al., 2019). Moreover, high-power optogenetic stimulation of CPs produced a similar effect (Hartmann et al., 2021).
Furthermore, calcium is involved in a plethora of cellular processes, ranging from cellular homeostasis to force generation (Berridge et al., 2003). This may explain the spread of calcium signals observed in CPs at basal conditions. The oscillatory nature of the CP calcium dynamic could be a way of reducing the threshold for activation of calcium-related pathways, such as regulation of gene expression (Dolmetsch et al., 1998). Regarding the importance of calcium in the contractile machinery of SMCs (Goulopoulou and Webb, 2014), a relationship between intracellular calcium in CPs and vasoactivity is possible: We observed that elevations of calcium in CPs were associated with constriction (Figure 3G, Figure 3—figure supplement 2), while drops in calcium were associated with dilations (Figure 5D, Figure 6—figure supplement 3). This suggests that CPs may be able to exert calcium-related forces on capillaries with possible impact on blood flow regulation. Just recently it could be shown that optogenetic stimulation of CPs could induce capillary constriction. These constrictions could be prevented by fasudil, an inhibitor of actomyosin–cytoskeleton regulatory rho-kinases (Hartmann et al., 2021). Since expression of smooth muscle actin in CPs seems absent or very low (Grant et al., 2019; Alarcon-Martinez et al., 2018), the type of contractile machinery in CPs remains to be elucidated. Further knowledge of the pericyte calcium-signaling toolkit (Hariharan et al., 2020) may help to understand the consequences of altered calcium signaling in pericytes in physiological and pathophysiological conditions.
Potassium is released in high amounts during the repolarization phase after action potential firing (Paulson and Newman, 1987) and can act as a potent vasomodulator (McCarron and Halpern, 1990). Potassium sensing by SMCs and capillary ECs via Kir2 channels has been previously described as a mechanism to increase local cerebral blood flow (Longden and Nelson, 2015; Longden et al., 2017; Filosa et al., 2006; Haddy et al., 2006). Furthermore, a modeling study showcases the capillary EC Kir2 channel as a sensor of neuronal activity and highlights its impact on potassium-mediated neurovascular communication (Moshkforoush et al., 2020). However, so far the role of CPs in this potassium-mediated vascular response remains elusive. Previous studies revealed a strong correlation between CP loss and dysregulated neurovascular coupling (Kisler et al., 2017; Nikolakopoulou et al., 2019). A recent study, using a pressurized retina preparation, intimates that junctional pericytes act as control elements in the potassium-mediated functional hyperemia by directing blood flow via branch-specific dilation to a site where a stimulus was evoked (Gonzales et al., 2020). Moreover, optogenetic stimulation of CPs was recently shown to induce capillary constriction (Hartmann et al., 2021), suggesting that CPs could be involved in neurovascular coupling.
In our study, we report that specific excitatory chemogenetic stimulation of neurons in vivo decreases calcium signals in all mural cells along a vascular branch and this was associated with an increase in vascular diameters including capillaries (Figure 5, Video 6). Comparable calcium signal reductions were found by odorant-stimulated neurons (Rungta et al., 2018) or during spreading depolarization (Khennouf et al., 2018). However, the threshold of neuronal excitation to cause a calcium response in CPs still needs to be further determined, since another study could not observe a calcium signal change in CPs in response to sensory whisker stimulation (Hill et al., 2015). We also demonstrate that CP calcium signals decrease in response to an acute rise of extracellular potassium concentration in acute brain slices (Figure 6A). Pharmacological inhibition of Kir2 and KATP channels inhibited the potassium evoked calcium drop, implying a role of these channels in silencing calcium activity in CPs (Figure 6B and C). Moreover, both in vivo hypoxia and blockade of mitochondrial respiration ex vivo, presumably leading to a reduced intracellular [ATP]/[ADP] ratio and likely elevated potassium levels due to neuronal depolarization, also resulted in cessation of CP calcium signals (Figure 6D and E), possibly as a result of Kir2 and KATP channel activation. High abundance of Kir2 and KATP channel transcripts were reported in CPs (Vanlandewijck et al., 2018; Bondjers et al., 2006). Interestingly, a recent study could show that KATP channel activation leads to a hyperpolarization in CPs in retinal preparations (Sancho et al., 2021).
However, Kir2 and KATP channels are also expressed in capillary ECs, which mediate a hyperpolarization in response to increases in extracellular potassium (Longden et al., 2017). Given that there is evidence of EC and CP gap-junctional coupling in the retinal vasculature (Wu et al., 2006; Kovacs-Oller et al., 2020; Ivanova et al., 2017; Ivanova et al., 2019) and the lack of cell-type specificity of pharmacological manipulations, we cannot exclude that the observed potassium evoked calcium drop in CPs could be secondary to changes in capillary ECs. However, still very little is known of how capillary ECs interact and modulate CP functions or vice versa. Gap-junction coupling of capillary ECs and CPs would allow for electrical interconnection between these cells, forming a vascular relay (Ivanova et al., 2019). Activation of Kir2 and KATP channels likely induces cellular hyperpolarization in this EC–CP capillary unit, which could be transmitted between and within CPs and ECs through gap-junctional coupling (Figure 7). Potassium sensing by CPs and ECs could potentiate the propagation of hyperpolarization from a site of elevated neuronal activity to upstream feeding arterioles.

Working model of potassium-induced calcium changes in capillary pericytes.
Neuronal activity leads to the release of potassium into the extracellular space.This rise in potassium [K+] activates Kir2.1 channels on capillary endothelial cells (EC) to induce a retrograde propagating hyperpolarization via gap-junctional coupling (Longden et al., 2017), which leads to upstream dilation of arteries. Elevated neuronal activity and a rise in extracellular potassium decrease calcium signaling in pericytes. Kir2.2 and KATP channels expressed on capillary pericytes (CPs) may induce a hyperpolarization in CPs as recently shown (Sancho et al., 2021). This hyperpolarization would inactivate TRPC and VGCC channels in CPs leading to a drop in calcium signals. Hypoxia or inhibition of respiratory metabolism reduce the [ATP]/[ADP] ratio thereby activating KATP and Kir2.2 channels to induce a hyperpolarization. By gap-junction coupling (Cx 37, 43) between ECs and CPs, the hyperpolarization of the EC–CP capillary unit could be transmitted retrogradely to induce upstream vascular responses.
Moreover, CPs are optimally positioned in the capillary bed (like cellular antennas of the EC–CP unit) to sense the microenvironment (Pfeiffer et al., 2021) and to amplify the hyperpolarization-mediated vascular response. Future studies using concurrent calcium imaging in ECs and CPs as well as cell-specific knock-out models are needed to gain more insights into the individual and the intercellular contribution of capillary ECs and CPs in potassium sensing and neurovascular coupling.
Materials and methods
Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
---|---|---|---|---|
Genetic reagent (M. musculus) | B6.Cg-Tg(Pdgfrb-CreERT2)6,096Rha/J | Jackson Laboratory | #029684RRID:IMSR_JAX:029684 | Pdgfrb-CreERT2 |
Genetic reagent (M. musculus) | Ai96 | Jackson Laboratory | #028866RRID:IMSR_JAX:024106 | GCaMP6s |
Recombinant DNA reagent | AAV2-hSYN-hM3D (Gq)-mCherry | VVF, UZH Zuerich | ID: v101 | |
Chemical compound, drug | Tamoxifen | Sigma-Aldrich | Cat. #: T5648 | 10 mg/ml |
Chemical compound, drug | Isoflurane | Piramal Healthcare | Attane | |
Chemical compound, drug | Texas Red Dextrane 70 kDa | Life Technologies | Cat. #: D-1864 | 2.5% |
Chemical compound, drug | Nimodipine | Tocris | Cat. #: 0600 | |
Chemical compound, drug | Endothelin-1 | Tocris | Cat. #: 1,160 | |
Chemical compound, drug | TTX citrate | Tocris | Cat. #: 1,069 | |
Chemical compound, drug | SKF96365 HCl | Tocris | Cat. #: 1,147 | |
Chemical compound, drug | PNU-37883A HCl | Tocris | Cat. #: 2095 | |
Chemical compound, drug | U46619 | Tocris | Cat. #: 1932 | |
Chemical compound, drug | Clozapine | Tocris | Cat. #: 0444 | |
Chemical compound, drug | UDP-Glucose 2Na+ | Sigma-Aldrich | Cat. #: 94,335 | |
Chemical compound, drug | ATP | Sigma-Aldrich | Cat. #: A2383 | |
Chemical compound, drug | DMSO | Sigma-Aldrich | Cat. #: 41,640 | |
Software, algorithm | R studio | RStudio Team (2020) | v.1.0.136 | http://www.rstudio.com/ |
Software, algorithm | GraphPad Prism | GraphPad Software La Jolla | V7.0RRID:SCR_002798 | |
Software, algorithm | Matlab | MathWorks | R2017bRRID:SCR_001622 | |
Software, algorithm | Matlab code CHIPS | https://ein-lab.github.io/ Barrett et al., 2018 | ||
Software, algorithm | ImageJ | http://imageJ.nih.gov/ij | 1.53 cRRID:SCR_003070 | |
Software, algorithm | ImageJ Vessel diameter plugin | PMID:20406650 | v.1.0 | |
Software, algorithm | ScanImage | Janelia Research Campus | R3.8.1RRID:SCR_014307 | |
Other | Inline heated perfusion cube | ALA Scientific Instruments | HPC-G | |
Other | Sapphire glass | UQG Optics | Ø 3 × 3 mm |
Animals
All animal experiments were approved by the local Cantonal Veterinary Office in Zürich (license ZH 169/17) and conformed to the guidelines of the Swiss Animal Protection Law, Swiss Veterinary Office, Canton of Zürich (Animal Welfare Act of 16 December 2005 and Animal Protection Ordinance of 23 April 2008). The following mice were interbred: B6.Cg-Tg(Pdgfrb-CreERT2)6,096Rha/J (Pdgfrb-CreERT2), The Jackson Laboratory (Stock #029684, Gerl et al., 2015) with Rosa26< LSL-GCaMP6s> (Ai96), The Jackson Laboratory (Stock #028866). Male and female offspring aged 2–9 months were used for experiments. The mice were given free access to water and food and were maintained under an inverted 12/12 hr light/dark cycle.
Experimental timeline
Request a detailed protocolTo induce GCaMP6s expression, Tamoxifen (Sigma-Aldrich, cat. no. T5648), dissolved in corn oil (10 mg/ml), was injected intraperitoneally in mice at a dose of 100 mg/kg on four consecutive days, 3 weeks before ex vivo or in vivo imaging. After cranial window implantation, mice were allowed to recover for 3 weeks prior to in vivo imaging, to ensure that all surgery-related inflammation had resolved.
Anesthesia
Request a detailed protocolFor surgery, animals were anesthetized with a mixture of fentanyl (0.05 mg/kg bodyweight; Sintenyl; Sintetica), midazolam (5 mg/kg bodyweight; Dormicum, Roche), and medetomidine (0.5 mg/kg bodyweight; Domitor; Orion Pharma), administered intraperitoneally. Anesthesia was maintained with midazolam (5 mg/kg bodyweight), injected subcutaneously 50 min after induction. To prevent hypoxemia, a face mask provided 300 ml/min of 100% oxygen. During two-photon imaging mice were anesthetized with 1.2% isoflurane (Attane; Piramal Healthcare, India) and supplied with 300 ml/min of 100% oxygen. Core temperature was kept constant at 37°C using a homeothermic heating blanket system (Harvard Apparatus) during all surgical and experimental procedures. The animal’s head was fixed in a stereotaxic frame (Kopf Instruments) and the eyes were kept moist with ointment (vitamin A eye cream; Bausch & Lomb).
Head-post implantation
Request a detailed protocolA bonding agent (Gluma Comfort Bond; Heraeus Kulzer) was applied to the cleaned skull and polymerized with a handheld blue light source (600 mW/cm2; Demetron LC). A custom-made aluminum head-post was connected with dental cement (Synergy D6 Flow; Coltene/Whaledent AG) to the bonding agent on the skull for later reproducible animal fixation in the microscope setup. The skin lesion was treated with antibiotic ointment (Neomycin, Cicatrex; Janssen-Cilag AG) and closed with acrylic glue (Histoacryl; B. Braun). After surgery, the animals were kept warm and were given analgesics (Temgesic [buprenorphine] 0.1 mg/kg bodyweight; Sintetica).
Virus injection and cranial window surgery
Request a detailed protocolA 4 × 4 mm craniotomy was performed above the somatosensory cortex using a dental drill (Bien-Air Dental), and for experiments requiring chemogenetics, adeno-associated virus (AAV) vectors were injected into the primary somatosensory cortex to achieve a localized chemogenetic receptor protein expression: 50 nl of AAV2-hSYN-hM3D (Gq)-mCherry (titer 1.02 × 1011 VG/ml; Viral Vector Core Facility [VVF], University of Zürich) at a cortical depth of 300 µm. Large vessels were avoided to prevent bleeding. A square Sapphire glass coverslip (3 × 3 mm, UQG Optics) was placed on the exposed dura mater and fixed to the skull with dental cement, according to published protocols (Holtmaat et al., 2009).
Behavior training for awake two-photon imaging
Request a detailed protocolOne week post-surgery, animals were handled multiple times a day for a week in order to get familiarized with the experimenter. Then, animals were adapted to the head fixation box by restraining them via the implanted head-post several times a day, with a gradual increase in restraint from seconds up to several minutes. After an extensive training period of 2–3 weeks, animals learned to sit still for the duration of a 45 min imaging session. Prior to an imaging session, 100 µl of a 2.5%, 70 kDa Texas Red Dextran (Life Technologies, cat. no. D-1864) was injected intravenously (iv.) into the tail vein to stain blood plasma. Mice were then let to recover from the iv. application and anesthesia for at least 1 hr before awake imaging was conducted.
Two-photon imaging
Request a detailed protocolTwo-photon imaging was performed using a custom-built two-photon laser scanning microscope (2PLSM) (Mayrhofer et al., 2015) with a tunable pulsed laser (MaiTai HP system, Spectra-Physics and Chameleon Discovery TPC, Coherent Inc) and equipped with a 20× (W Plan-Apochromat 20 x/1.0 NA, Zeiss) or 25× (W Plan-Apochromat 25 x/1.05 NA, Olympus) water-immersion objective. During in vivo measurements, the animals were head-fixed and kept under anesthesia as described above. To visualize the vasculature Texas Red Dextran (2.5% w/v, 70,000 mw, 50 µl, Life Technologies, cat. no. D-1864) was injected intravenously. GCaMP6s and Texas Red were excited at 940 nm, and emission was detected with GaAsP photomultiplier modules (Hamamatsu Photonics) fitted with 535/50 nm and 607/70 nm band-pass filters and separated by a 560 nm dichroic mirror (BrightLine; Semrock). Control of microscope laser scanning was achieved with a customized version of ScanImage (r3.8.1; Janelia Research Campus; Pologruto et al., 2003).
At the beginning of each imaging session, z-stacks of the area of interest were recorded to identify the branch order of individual capillary segments. Once capillaries with a branch order of >4 were identified, a high resolution (512 × 512 pixels, 0.74 Hz) image was collected for reference and then baseline images (128 × 128 pixels; 11.84 Hz) were collected over a 90 s period with zoom factors ranging from 10 to 19. Multiple imaging sessions were conducted on different days for each animal. Calcium imaging of SMCs and EPs was limited to 25 min after isoflurane anesthesia induction.
Acute brain slice preparation
Request a detailed protocolPrior to slicing, 100 µl of a 2.5%, 70 kDa Texas Red Dextran (Life Technologies, cat. no. D-1864) was injected intravenously into the tail vein to stain blood plasma. Mice were euthanized by decapitation after deep anesthesia with isoflurane. The brain was extracted from the skull in ice-cold cutting solution consisting of 65 mM NaCl, 2.5 mM KCl, 0.5 mM CaCl2, 7 mM MgCl2, 25 mM glucose, 105 mM sucrose, 25 mM NaHCO3, and 2.5 mM Na2HPO4. Slices of 300 µm thickness were cut using a Microm HM 650 V vibratome. Slices were immediately transferred into artificial cerebrospinal fluid (aCSF) consisting of 126 mM NaCl, 3 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 25 mM glucose, 26 mM NaHCO3, and 1.25 mM NaH2PO4, at 34°C. After a recovery phase of 1 hr, slices were used for imaging. All solutions were continuously gassed with 95% O2, 5% CO2 and were prepared on the day of experiment.
Pharmacology in brain slices
Request a detailed protocolSlices were imaged at 34°C in the same aCSF that was used for recovery after cutting. Solutions were infused by gravity into the slice chamber at a rate of 1.5 ml/min with an In-line Heated Perfusion Cube (HPC-G; ALA Scientific Instruments).
Nimodipine (Cat. No. 0600), U46619 (Cat. No. 1932), Endothelin-1 (Cat. No. 1160), TTX citrate (Cat. No. 1069), SKF96365 (Cat. No. 1147), and PNU-37883A HCl (Cat. No. 2095) were obtained from Tocris Bioscience (USA). UDP-Glucose (Cat. No. 94335), ATP (Cat. No. A2383), and all salts were obtained from Sigma-Aldrich.
Nimodipine, SKF96365, and PNU-37883A HCl were dissolved in DMSO, and Endothelin-1, UDP-Glucose, ATP and TTX citrate were dissolved in water as stock solutions at the highest possible molarity, stored at –20°C and were diluted to the desired concentration in aCSF prior to use.
For aCSF with increased potassium concentrations, the sodium-ion quantity was adapted to maintain the osmolarity of the solution constant.
Acute slice electrophysiology
Request a detailed protocolCoronal slices were transferred to a submerged recording chamber (RC-26, Warner Instruments, Hamden, CT) mounted on an upright Olympus microscope (BX61WI) equipped with differential interference contrast. aCSF was perfused and maintained at 34°C by an inline solution heater (TC-344C, Warner Instruments). Cells were visualized using a 40× water-immersion objective (LUMPlanFI/IR 40 x/0.80 W, Olympus). Somatic whole-cell recordings were performed from cortical layer 2/3 excitatory neurons using 3–4 MΩ glass pipettes filled with (in mM): 135 K-gluconate, 4 KCl, 2 NaCl, 10 HEPES, 4 EGTA, 4 Mg-ATP, 0.3 Na-GTP (pH 7.2, adjusted with KOH). Intrinsic properties of the neurons were measured in current clamp mode (in the absence of synaptic blockers) and depolarizing current injection steps of 500 ms duration were used to elicit action potentials. Tetrodotoxin TTX (1 µM) was added in the bathing solution to block neuronal spiking activity. Recordings were acquired with a Multiclamp 700B amplifier (Axon Instruments, Union City, CA), low-pass filtered at 2 kHz, digitized at 20 kHz (using Digidata 1,550B, Axon Instruments), and stored to disk using pClamp10 software (Molecular Devices, Sunnyvale, CA). Data analysis was performed offline using Clampfit 10.6 software (Molecular Device).
Hypoxia pprotocol
Request a detailed protocolTo induce an acute hypoxic insult to the mice, the oxygen supply was transiently replaced with 2.0 l/min 100% nitrogen for 45 s.
DREADD activation
Request a detailed protocolTo visualize and assess the area of hM3D(Gq)-mCherry transduced neurons, the laser wavelength was changed to 990 nm for overview imaging. Ca2+ recordings were performed at a wavelength of 940 nm. hM3D(Gq) DREADD transduced neurons were activated with 30 µg/kg bodyweight clozapine (Tocris, Cat. No. 0444) (Jendryka et al., 2019) in saline (0.9% w/v) via a tail vein injection during image acquisition.
Quantification and statistical analysis
Request a detailed protocolImage analysis was performed using ImageJ and a custom-designed image processing toolbox, Cellular and Hemodynamic Image Processing Suite (CHIPS, Barrett et al., 2018), based on MATLAB (R2017b, MathWorks). For each field of view, all images were spectrally unmixed to reduce potential bleed-through between imaging channels and aligned using a 2D convolution engine to account for motion and x–y drift in time. Background noise was defined as the bottom first percentile pixel value in each frame and was subtracted from every pixel. Regions of interest (ROIs) were selected by combining two distinct methods in CHIPS: hand-selection of cell bodies as well as whole cell, and automated ROI detection with an activity-based algorithm (Ellefsen et al., 2014), both using anatomical images (128 × 128 pixel). A 2D spatial Gaussian filter (σxy = 2 µm) and a temporal moving average filter (width = 1 s) were applied to all images to reduce noise. A moving threshold for each pixel was defined in the filtered stack as the mean intensity plus seven times the standard deviation of the same pixel during the preceding 2.46 s. Using this sliding box-car approach, active pixels were identified as those that exceeded the threshold. Active pixels were grouped in space (radius = 2 µm) and time (width = 1 s). Resulting ROIs with an area smaller than 4 µm2 were considered to be noise and were excluded. We then combined the previously hand-selected ROIs into a single mask by subtracting the soma ROI from the whole cell territory ROIs, thereby leaving a mask of pericytes without soma. We then multiplied this 2D mask with each frame of the 3D mask obtained from the automated ROI detection in order to obtain a mask of ROIs within the cell territory and outside of the soma. We then extracted traces from two sets of ROIs for each image: the hand-selected soma ROIs and the adjusted 3D activity mask. The minimum distance from each activity ROI to the nearest soma was defined as the shortest distance between ROI edges. The signal vector (df/f) from each ROI was calculated using the mean intensity between 0.4 s and 4 s as baseline. Short, fast peaks were identified by applying a digital band-pass filter with passband frequencies (f1 = 0.025 Hz and f2 = 0.2 Hz) before running the MATLAB findpeaks function. Noise peaks, due to motion or inflow of high-fluorescent particles, were manually removed.
Line scan acquisitions for vessel diameter were also analyzed using the custom-designed image processing toolbox for MATLAB, employing implemented methods described earlier (Kim et al., 2012; Drew et al., 2010; Gao and Drew, 2014). Vessel diameters for clozapine and hypoxia images were determined from time-averaged (20 s) stacks using the ImageJ vessel diameter plugin (Fischer et al., 2010).
Statistics for in vivo data was performed in RStudio (version 1.0.136) using the lme4 package for linear mixed-effects models (Bates et al., 2015). For fixed effects, we used the experimental condition (with/without drug) or cell type. For random effects, we had intercepts for individual animals and cells. Likelihood ratio tests comparing models with fixed effects against models without fixed effects were used to determine the model with the best fit while accounting for the different degrees of freedom. Visual inspection of residual plots did not reveal any obvious deviations from homoscedasticity or normality. All data were reported and plotted as uncorrected means. Frequencies are reported as signals/min ± SD, amplitudes as df/f ± SD and durations as s ± SD. p-values for different parameter comparisons were obtained using the multcomp package with Tukey’s post hoc tests.
Graphics and statistical analyses of ex vivo experiments were performed using GraphPad Prism (version 7.00; GraphPad Software, La Jolla, CA). Datasets were tested with a Shapiro–Wilk normality test for Gaussian distribution. If one or both of the datasets passed the normality test, a two-tailed paired t-test was carried out; otherwise, a Wilcoxon matched-pairs signed rank test was chosen. For multiple comparisons, a one-way ANOVA was performed, using the Greenhouse–Geisser correction for sphericity and Tukey’s post hoc tests for group comparisons. All statistical tests used to evaluate significance are indicated in the Figure legends along with the p-values. Values for area under the curve (auc) were calculated from 20 s windows of normalized traces for both baseline and drug responses considering the whole cell (soma and processes). For all in vivo and ex vivo experiments, N gives the number of animals and n is the number of cells. Statistical significances are highlighted based on: *p<0.05, **p<0.01, ***p<0.001.
Data and Software Availability
Request a detailed protocolThe CHIPS toolbox for MATLAB is freely available on GitHub (https://ein-lab.github.io/; Barrett et al., 2020; Barrett et al., 2018). The source code for calcium signal detection with CHIPS is available as source code file.
Data availability
All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2, 3, 4, 5 and 6.
References
-
What is a pericyte?Journal of Cerebral Blood Flow and Metabolism 36:451–455.https://doi.org/10.1177/0271678X15610340
-
Fitting Linear Mixed-Effects Models Using lme4Journal of Statistical Software 67:1–48.https://doi.org/10.18637/jss.v067.i01
-
Calcium signalling: Dynamics, homeostasis and remodellingNature Reviews. Molecular Cell Biology 4:517–529.https://doi.org/10.1038/nrm1155
-
Pericyte structural remodeling in cerebrovascular health and homeostasisFrontiers in Aging Neuroscience 10:210.https://doi.org/10.3389/fnagi.2018.00210
-
BookIn vivo optical imaging and manipulation of brain pericytesIn: BirbrairI A, editors. Biology of Pericytes – Recent Advances. Cham: Springer International Publishing. pp. 1–37.https://doi.org/10.1007/978-3-030-62129-2_1
-
A review of string vessels or collapsed, empty basement membrane tubesJournal of Alzheimer’s Disease 21:725–739.https://doi.org/10.3233/JAD-2010-100219
-
Pericyte bridges in homeostasis and hyperglycemiaDiabetes 69:1503–1517.https://doi.org/10.2337/db19-0471
-
Activation of atp-dependent k+ channels by hypoxia in smooth muscle cells isolated from the pig coronary arteryThe Journal of Physiology 483 ( Pt 1):29–39.https://doi.org/10.1113/jphysiol.1995.sp020565
-
Endothelin-1 as a mediator of endothelial cell-pericyte interactions in bovine brain capillariesJournal of Cerebral Blood Flow and Metabolism 17:464–469.https://doi.org/10.1097/00004647-199704000-00012
-
Rapid determination of particle velocity from space-time images using the radon transformJournal of Computational Neuroscience 29:5–11.https://doi.org/10.1007/s10827-009-0159-1
-
Transient receptor potential channels in the vasculaturePhysiological Reviews 95:645–690.https://doi.org/10.1152/physrev.00026.2014
-
Local potassium signaling couples neuronal activity to vasodilation in the brainNature Neuroscience 9:1397–1403.https://doi.org/10.1038/nn1779
-
Measurement of meningeal blood vessel diameter in vivo with a plug-in for ImagejMicrovascular Research 80:258–266.https://doi.org/10.1016/j.mvr.2010.04.004
-
Endothelial-mural cell signaling in vascular development and angiogenesisArteriosclerosis, Thrombosis, and Vascular Biology 29:630–638.https://doi.org/10.1161/ATVBAHA.107.161521
-
Determination of vessel cross-sectional area by thresholding in radon spaceJournal of Cerebral Blood Flow and Metabolism 34:1180–1187.https://doi.org/10.1038/jcbfm.2014.67
-
Inducible glomerular erythropoietin production in the adult kidneyKidney International 88:1345–1355.https://doi.org/10.1038/ki.2015.274
-
Organizational hierarchy and structural diversity of microvascular pericytes in adult mouse cortexJournal of Cerebral Blood Flow and Metabolism 39:411–425.https://doi.org/10.1177/0271678X17732229
-
Cellular control of brain capillary blood flow: In vivo imaging VeritasTrends in Neurosciences 42:528–536.https://doi.org/10.1016/j.tins.2019.05.009
-
Role of potassium in regulating blood flow and blood pressureAmerican Journal of Physiology-Regulatory, Integrative and Comparative Physiology 290:R546.https://doi.org/10.1152/ajpregu.00491.2005
-
Signalling and pharmacological properties of the P2Y receptorActa Physiologica 199:149–160.https://doi.org/10.1111/j.1748-1716.2010.02116.x
-
The ion channel and GPCR toolkit of brain capillary pericytesFrontiers in Cellular Neuroscience 14:601324.https://doi.org/10.3389/fncel.2020.601324
-
A murine toolbox for imaging the neurovascular unitMicrocirculation 22:168–182.https://doi.org/10.1111/micc.12176
-
Nature NeuroscienceBrain Capillary Pericytes Exert a Substantial but Slow Influence on Blood Flow Vol. 24:633–645.https://doi.org/10.1038/s41593-020-00793-2
-
Calcium signaling in smooth muscleCold Spring Harbor Perspectives in Biology 3:a004549.https://doi.org/10.1101/cshperspect.a004549
-
Vasomotion and flowmotion: Physiological mechanisms and clinical evidenceVascular Medicine Review 1:101–112.https://doi.org/10.1177/1358836X9000100202
-
Domain-specific distribution of gap junctions defines cellular coupling to establish a vascular relay in the retinaThe Journal of Comparative Neurology 527:2675–2693.https://doi.org/10.1002/cne.24699
-
Pericyte degeneration leads to neurovascular uncoupling and limits oxygen supply to brainNature Neuroscience 20:406–416.https://doi.org/10.1038/nn.4489
-
Regulation of cell-to-cell communication mediated by astrocytic atp in the CNSPurinergic Signalling 1:211–217.https://doi.org/10.1007/s11302-005-6321-y
-
Experimental chronic cerebral hypoperfusion results in decreased pericyte coverage and increased blood-brain barrier permeability in the corpus callosumJournal of Cerebral Blood Flow and Metabolism 39:240–250.https://doi.org/10.1177/0271678X17743670
-
Ion channel networks in the control of cerebral blood flowJournal of Cerebral Blood Flow and Metabolism 36:492–512.https://doi.org/10.1177/0271678X15616138
-
Design and performance of an ultra-flexible two-photon microscope for in vivo researchBiomedical Optics Express 6:4228–4237.https://doi.org/10.1364/BOE.6.004228
-
Potassium dilates rat cerebral arteries by two independent mechanismsAmerican Journal of Physiology-Heart and Circulatory Physiology 259:H902.https://doi.org/10.1152/ajpheart.1990.259.3.H902
-
Invited review: Significance of spatial and temporal heterogeneity of calcium transients in smooth muscleJournal of Applied Physiology 91:488–496.https://doi.org/10.1152/jappl.2001.91.1.488
-
Diverse mechanisms regulating brain energy supply at the capillary levelCurrent Opinion in Neurobiology 69:41–50.https://doi.org/10.1016/j.conb.2020.12.010
-
Scanimage: Flexible software for operating laser scanning microscopesBiomedical Engineering Online 2:13.https://doi.org/10.1186/1475-925X-2-13
-
A pharmacological model of ischemia in the hippocampal sliceNeuroscience Letters 119:175–178.https://doi.org/10.1016/0304-3940(90)90827-v
-
Effects of anesthesia on cerebral blood flow, metabolism, and neuroprotectionJournal of Cerebral Blood Flow and Metabolism 38:2192–2208.https://doi.org/10.1177/0271678X18789273
-
Extracellular potassium in the mammalian central nervous systemAnnual Review of Physiology 41:159–177.https://doi.org/10.1146/annurev.ph.41.030179.001111
-
Pericyte-expressed tie2 controls angiogenesis and vessel maturationNature Communications 8:16106.https://doi.org/10.1038/ncomms16106
-
Atp-sensitive potassium channels and their physiological and pathophysiological rolesComprehensive Physiology 8:1463–1511.https://doi.org/10.1002/cphy.c170048
-
Brain microvascular pericytes in vascular cognitive impairment and dementiaFrontiers in Aging Neuroscience 12:80.https://doi.org/10.3389/fnagi.2020.00080
-
Vascular smooth muscle cell signaling mechanisms for contraction to angiotensin ii and endothelin-1Journal of the American Society of Hypertension 3:84–95.https://doi.org/10.1016/j.jash.2008.09.002
-
Pericytes in primary familial brain calcificationAdvances in Experimental Medicine and Biology 1147:247–264.https://doi.org/10.1007/978-3-030-16908-4_11
-
Der Feinere Bau der BlutcapillarenZeitschrift Für Anatomie Und Entwicklungsgeschichte 68:29–109.https://doi.org/10.1007/BF02593544
Decision letter
-
Mark T NelsonReviewing Editor; University of Vermont, United States
-
Kenton J SwartzSenior Editor; National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
-
Mark T NelsonReviewer; University of Vermont, United States
-
Jaime GrutzendlerReviewer; Yale University, United States
-
Andy ShihReviewer
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Acceptance summary:
This is a rigorous and comprehensive study of brain vascular mural cell physiology using state of the art imaging , electrophysiology and pharmacology. This study will be of interest to neuroscientists and vascular biologists.
Decision letter after peer review:
[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]
Thank you for submitting your work entitled "Distinct signatures of calcium activity in brain pericytes" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Andy Shih (Reviewer #2); Jaime Grutzendler (Reviewer #3).
Our decision has been reached after excessive consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
The reviewers felt that the calcium signaling measurements in vivo were of significant interest. However, there were many technical concerns including the viability of the brain slice preparation and interpretation of potassium channel drug effects on multiple cell types. We felt that these issues would take considerable time, and would likely change the conclusions of the paper. eLife policy indicates that revisions should be limited to those that can be reasonably accomplished in two months. We would be willing to consider a new manuscript that addressed the concerns of the reviewers.
Reviewer #1:
Summary:
In their manuscript, Glück et al., seek to define Ca2+ signaling characteristics in brain pericytes, focusing on differences in Ca2+ events between arteriole-proximate ensheathing pericytes (EPs) and more distal pericytes, which they refer to as capillary pericytes (CPs). Using 2-photon imaging in an anesthetized mouse cranial window model, they report distinct differences in Ca2+ event frequency, amplitude and duration between EPs and CPs. They further describe a clear difference in the Ca2+-contraction relationship between these two different pericyte populations, reporting that changes in Ca2+ in EPs are rapidly followed by changes in vessel diameter, a response pattern that mirrors that in arteriolar smooth muscle cells (SMCs); in contrast, CPs showed no change in diameter in response to Ca2+ events. Subsequent experiments in acute brain slices suggested differences in Ca2+ signal persistence ex vivo versus in vivo and provided evidence that vasomodulating Gq-protein-coupled receptor (GqPCR) agonists are capable of driving Ca2+ events in CPs. Pharmacological interventions suggested that these events are dependent on Ca2+ influx via TRPC (transient receptor potential canonical) channels, but independent of ongoing neuronal activity. Their data further suggest that the L-type voltage-gated Ca2+ channel may also mediate Ca2+ influx, although this is not likely a major influx pathway as the selective blocker nimodipine had no effect on Ca2+ events in processes and only a very modest effect on such events in somata. Finally, on the basis of pharmacological approaches in brain slices, including DREADD-based experiments, direct application of K+ and ischemia/hypoxia-mimetic conditions (sodium azide/100% N2 exposure), they suggest that K+ released in response to neuronal stimulation causes a decrease in Ca2+ events in CPs, and that Kir and KATP channels mediate this effect.
General comments/questions
Data are dumped wholesale in parenthetic expressions together with statistical information and figure call-outs, making it difficult to assess relationships. Such data-rich sentences should be re-written to more closely juxtapose data and associated conditions for more facile comparison. For example (lines 201-203):
Before: "Comparing the signal frequency between in vivo and ex vivo, there was a more than twofold reduction in calcium activity in both somata and processes (in vivo, S: 7.8 (2.5) and P: 18.9 (9.4); ex vivo, S: 3.1 (1.6) and P: 7.3 (5.6). Values are mean (SD) signals/min. Figure 3A)."
After: "A comparison of calcium activity showed that calcium signal frequency, expressed as mean ± standard deviation (SD), was reduced more than 2-fold ex vivo compared with that in vivo in both processes (7.3 ± 5.6 vs. 18.9 ± 9.4) and somata (3.1 ± 1.6 vs. 7.8 ± 2.5) (Figure 3A)."
Distinct basal calcium transients of mural cells in vivo
Lines 126-7: "Basal calcium signals in SMCs and EPs 126 were usually only visible during the first 30 minutes of anesthesia." When exactly during this 30-min period were Ca2+ events in EPs recorded? Immediately after delivering anesthesia (i.e., before anesthesia effects on signaling manifested)? Some time during the 30-minute period? At the end of the 30-minute period? Measurements made immediately after delivering anesthesia, though more difficult to obtain, might provide physiologically meaningful information, but the only thing learned from responses measured during or at the end of the 30-minute period is how anesthesia affects Ca2+ signaling in EPs.
Persisting calcium signals in CPs ex vivo
Line 204-5: In many cases (Figure 3B), arteries and arterioles were either collapsed due to loss of tone or intraluminal pressure in the slice preparation or were constricted by possibly dead SMCs or EPs (no detectable calcium signal). These observations are disturbing signs of possible tissue injury and raise questions about the validity of data obtained using acute brain slice preparations. Faced with these observations, the authors should confirm cell viability in brain slice preparations after all pharmacological interventions that reduce Ca2+ signaling (e.g., hM3D(Gq)-DREADD/clozapine, SKF) to rule out the possibility that observed decreases in Ca2+ events reflect the trivial case of signal rundown in dying cells.
Line 226-7: "…massive calcium response in CPs when U46619 was included in the superfusate (Figure 3D, 226 Video 5). This overt calcium response in CPs was accompanied by cytoplasmic extrusions, suggesting that U46619 might be toxic for pericytes." Although U46610 can have harsh effects, including pericyte blebbing, the severity of U46610 effects described here suggest that procedures used to isolate, prepare and/or maintain brain slices increase the vulnerability of the preparation to potentially toxic stimuli.
Line 233: Header for Figure 3 legend (Calcium dynamics in the absence of blood flow ex vivo is more affected in EPs than CPs) implies that differences in EP Ca2+ dynamics between the in vivo and ex vivo setting is caused by the absence of blood flow in the latter preparation. This is an overstatement; no data are provided to support a direct relationship between blood flow and EP Ca2+ dynamics.
Figure 3 (all): Are these all z-stacks/z-projections of vessels? If so, this should be indicated in the figure and/or figure legend. The appearance of pinching or narrowing could be due to vessel orientation relative to the imaging plane.
CP calcium events are evoked by vasomodulators
Line 279: "Application of Nimodipine (100 μM)…" What is the rationale for using 100 μM nimodipine? This is an astonishingly high concentration for a dihydropyridine-even in a brain slice. A rationale is similarly required for SKF, which is also normally used at considerably lower concentrations than 100 μM.
Lines 288-9: "We applied tetrodotoxin (TTX), a voltage-gated sodium channel blocker, to dampen neuronal activity (Zonta et al., 2003)….CP calcium signals were not affected" The absence of an effect of TTX could simply indicate that cells in the brain slice preparation are unresponsive because they are dead or dying. The viability of cells in the prep needs to be verified.
Potassium released by neuronal stimulation leads to a calcium signal drop in CPs
Lines 322-4: "We used chemogenetics (Roth, 2016), in which neurons – in this case transduced to express hM3D(Gq)-DREADD – were activated with 30 μg/kg iv. clozapine." Transduction procedure needs to be described in Methods, and possible toxicity associated with it needs to be tested by assessing cell viability before and after transduction. In the absence of these latter control experiments, the possibility that the loss of Ca2+ signaling results from toxic effects of clozapine or toxicity of the transduction procedure itself cannot be ruled out.
Lines 332-338: Based on the set up for this section ("potassium is sensed by SMCs, causing the suppression of calcium oscillations"), the authors seem to be operating under the assumption that 10 mM K+ and BaCl2 are acting directly on mural cells-CPs in the current context. This is an incredibly bold assumption given that effects of 10 mM K+ and BaCl2 could be indirect through hyperpolarization of endothelial cells, which express Kir2.1 channels that are robustly activated by extracellular K+ and are blocked by 100 μM BaCl2.
Effects of KATP channel blockers could similarly be indirect through actions on endothelial cells. Although RNA seq data would suggest a large role for these channels in pericytes, further confirmation of their function should be provided, such as activation with pinacidil and block of the KATP channel during metabolic changes.
Overall
Using predominantly an ex vivo acute brain slice preparation, the authors report differences in Ca2+ signaling signatures between EPs and CPs, lending additional experimental weight to the growing body of evidence that arteriole-proximate and more distal pericytes are functionally distinct. They describe differential effects of anesthesia on basal Ca2+ signaling in EPs (gradually eliminated) and CPs (largely unaffected) and show that Ca2+ signaling in EPs is reduced in brain slices. They further intimate (e.g., see Abstract, Figure 3 legend) that Ca2+ signaling in EPs requires blood flow; however, their data do not support this conclusion since flow effects on Ca2+ events were not tested directly. Importantly, the viability of cells in brain slices was not assessed following pharmacological manipulations that decreased Ca2+ signaling. In the absence of such control experiments, it is not possible to rule out tissue injury or improper slice maintenance as the cause of such decreases, especially given evidence for dead SMCs and EPs and unusually severe effects of U46619. Accordingly, some of the more compelling data from ex vivo pharmacological manipulations are suspect. The transient nature of EP Ca2+ responses recorded by 2-photon microscopy in cranial window model mice, reflecting signal dampening associated with the onset of anesthesia, similarly cast doubt on the significance of these in vivo findings. None of their data, including chemogenetic activation of hM3D(Gq)-DREADD-expressing neurons and pharmacological manipulations involving K+ and BaCl2, support the specific conclusion that neuronal K+ release causes the decrease/loss of pericyte calcium signaling. Importantly, pharmacological manipulations of Kir2 and KATP channels do not exclude endothelial cell contributions-especially given the importance of endothelial cell Kir2 in neurovascular coupling. More generally, the interconnected electrical properties of the microvasculature are not discussed or accounted for within the overall experimental design.
Reviewer #2:
This is an exciting and timely study to uncover the mechanisms and physiological relevance of pericyte calcium signaling. By using pericyte specific calcium sensing-reporter mice (Pdgfrbeta-CreERT2; GCaMP6s (Ai96-flox)), Gluck et al., identify distinct Ca signatures within the heterogeneous populations of pericytes along the brain vasculature including smooth muscle cells (SMCs), ensheathing pericytes (EPs; found along the branch orders 1-4), and capillary pericytes (CPs; found along > 4th order). While SMCs and EPs have synchronous calcium oscillations that are expectedly related to vessel diameter changes, CPs have asynchronous calcium signaling in somata and processes and there is no correlation with vessel diameter changes in the healthy anesthetized animal. This has been shown in some prior studies, but the current manuscript delves far deeper into the basis of pericyte calcium signals.
To pharmacologically test how calcium signals are regulated in pericytes, they moved to ex vivo, brain slice experiments. They show that blocking VGCC and TRPC reduces CP calcium signals. Activating neurons using chemogenetics reduced CP calcium signals in vivo. KCl, to mimic neuronal activity, also reduced CP calcium and this effect is attenuated by blocking Kir2 and Kir6.1 ex vivo. Creating hypoxia-like conditions (in vivo: N2 or ex vivo: sodium azide) reduced calcium signals in CPs. Conversely, a number of vasoactive substances known to constrict arterioles lead to large calcium influxes in CPs, including ET-1, ATP and UDP-glucose.
These studies pave new roads to understanding how pericytes are involved in NV coupling, and blood flow control. The approach is highly innovative and the in vivo-ex vivo approach is a major strength. However, there is a missing link that makes it difficult to understand the physiological relevance of the CP calcium signals as currently presented (comments 1 and 2). If addressed, this would substantially increase the impact of the work.
1) It is not clear how calcium signaling in CPs is related to electrical conductance along the capillaries leading to arteriole diameter change. For example, how do CP calcium changes relate to calcium in SMCs and EPs (and/or arteriolar diameter) on the same vascular tree during NV coupling in vivo? How does CP calcium levels relate to endothelial activity? Is it possible to block CP K+ signaling in vivo to test the hypothesis that upstream arteriolar dilation will be decreased? Perhaps the relatively selective expression of KATP in CPs can be used to leverage an in vivo pharmacological experiment. It is not necessary to address all these questions, but some additional data to link CP calcium to arteriolar dilation would strengthen the paper and substantiate the compelling model in Figure 6.
2) Local regulation of capillary flow: The presumption is that the sustained calcium changes evoked by vasomediators (ET-1, ATP) and neural activity are affecting local capillary diameter. This would be helpful to know because past studies have not been as rigorous in reporting vascular branch orders, leaving some ambiguity as the authors note. The data would help us understand whether CPs are autonomously able to regulate capillary diameter. This is important even if the stimulation becomes supra-physiological, as it clarifies whether CPs are a logical target for constriction in pathologies such as stroke, AD. Curiously, some of the data also show a slower component of calcium modulation can occur on the scale of tens of second to minutes (traces of Figure 5). Are these slow modulations related to capillary diameter? Similarly, how does N2-induced hypoxia result in vascular diameter changes along the various pericyte territories in relation to Ca signal changes?
3) Neural activity: DREADDs are gated by clozapine-N-oxide, rather than clozapine. It is correct that clozapine was used in this study, instead of its less bioactive counterpart? If so, a control non-DREADD expressing group might be needed to verify the intended goal of stimulating local neurons in vivo, as opposed to broad actions of clozapine. Further, it is suggested that basal neuronal activity in slice does not regulate CP calcium fluctuations because TTX has no effect. However, neuronal activity was not actually measured.
4) In addition to a Non-DREADD expressing control, some additional controls would increase rigor. To ensure that the fluctuation in calcium dynamics in pericytes is not influenced by animal motion or passing RBCs in vivo, comparison to a general fluorescent reporter like GFP would be helpful. Vehicle control experiments are not described for the ex vivo pharmacological experiments. If these were performed, it would be good to show.
Reviewer #3:
The manuscript by Glück et al., examined the patterns of calcium transients in vascular mural cells in the live mouse brain and brain slices using two photon imaging of calcium sensor mice, in combination with pharmacological and chemogenetic manipulations. They found that pericytes exhibit distinct calcium dynamics compared to vascular smooth muscle cells (vSMCs). Pericytes in vivo have compartmentalized, irregular, higher-frequency calcium transients, which are not correlated with vessel diameter changes, whereas calcium fluctuation of vSMCs are highly correlated with changes in vessel diameter. In ex vivo preparations, pericytes retained their spontaneous calcium properties in contrast to its loss in SMCs. Combined with chemogenetics in vivo and pharmacological manipulation in slices, the authors demonstrated neuronal activity can decrease calcium response in pericytes by elevating extracellular potassium, which is mediated by Kir2.2, Kir6.1. Energetics state, and hypoxia challenge can also affect pericyte Ca2+ activation through KATP channel.
Overall, this paper is technically rigorous and reproduces the majority of in vivo findings by recent publications (PMIDs: 26119027, 29937277). It also provides a more granular investigation of Ca2+ transients in pericytes (comparing soma and processes) and the pharmacological investigation of potassium channels and ATP in slices is also novel.
1) The main criticism we have about this paper is the nomenclature used to define the different mural cells which continues to confuse the field. Why use the title "signatures of calcium in brain pericytes" instead of simply saying in mural cells? This is surprising given that their data completely parallels data showing that SMCs and pericytes are distinct cells types. They introduce again the confusing term "ensheathing pericytes" despite their finding that EP essentially have identically characteristics to vSMCs (near identical calcium properties, contractility, expression of aSMA, circumferential anatomy, lack of compartmentalized Ca2+ signals and others) as well as findings from other groups including transcriptome data showing no such thing as subpopulations of pericytes (PMIDs: 29443965, 30129931) and data from uptake of small fluorescent molecules specifically by CP but not other mural cells (PMID: 28504673). We think it would be critical for this paper to clarify the field rather than continue to confuse the nomenclature and cite historical 100 old anatomical studies to justify it. Otherwise it is essential that they produce functional, structural and genetic evidence that there is such thing as a distinct ensheathing cell that is closer to pericytes than to vSMCs to justify the name. The logical terminology would be to call them "terminal vSMCs" or something similar to account for the minor morphological difference these cells have compared to the slightly more proximal vSMCs.
2) Although not essential for this project, it would have been nice to obtain concurrent endothelial and pericyte Ca recordings (using RCaMP and GCaMP sensors) to really understand the role of CP Ca2+. Is it similar to the described hyperpolarization propagation in endothelium? (PMID: 28319610), are there any interactions between CP and endothelium?
3) They should also comment on the fact that it is very difficult to target pericytes without having impact on other cell types by pharmacological manipulation. Only Kir6.6 (Kcnj8) blocking provides pericyte specificity, since BaCl2 can also affect endothelial Kir channels (PMID: 26840527), which may indirectly change pericytes calcium.
4) Is there any directionality in the propagation of the Ca transients in pericytes in vivo?
5) Do the microdomains of Ca in vitro become more synchronized between processes and Soma. If so, could this mean that in the in vitro prep the microenvironment around pericytes is more homogeneous than in vivo, thereby causing more homogeneity of Ca?
6) In their figure 6 model there is no mention of possible pericyte to pericyte or pericyte to SMC communication. This would be worth considering as these cells are likely to be gap junction coupled.
https://doi.org/10.7554/eLife.70591.sa1Author response
[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]
Reviewer #1:
Summary:
In their manuscript, Glück et al., seek to define Ca2+ signaling characteristics in brain pericytes, focusing on differences in Ca2+ events between arteriole-proximate ensheathing pericytes (EPs) and more distal pericytes, which they refer to as capillary pericytes (CPs). Using 2-photon imaging in an anesthetized mouse cranial window model, they report distinct differences in Ca2+ event frequency, amplitude and duration between EPs and CPs. They further describe a clear difference in the Ca2+-contraction relationship between these two different pericyte populations, reporting that changes in Ca2+ in EPs are rapidly followed by changes in vessel diameter, a response pattern that mirrors that in arteriolar smooth muscle cells (SMCs); in contrast, CPs showed no change in diameter in response to Ca2+ events. Subsequent experiments in acute brain slices suggested differences in Ca2+ signal persistence ex vivo versus in vivo and provided evidence that vasomodulating Gq-protein-coupled receptor (GqPCR) agonists are capable of driving Ca2+ events in CPs. Pharmacological interventions suggested that these events are dependent on Ca2+ influx via TRPC (transient receptor potential canonical) channels, but independent of ongoing neuronal activity. Their data further suggest that the L-type voltage-gated Ca2+ channel may also mediate Ca2+ influx, although this is not likely a major influx pathway as the selective blocker nimodipine had no effect on Ca2+ events in processes and only a very modest effect on such events in somata. Finally, on the basis of pharmacological approaches in brain slices, including DREADD-based experiments, direct application of K+ and ischemia/hypoxia-mimetic conditions (sodium azide/100% N2 exposure), they suggest that K+ released in response to neuronal stimulation causes a decrease in Ca2+ events in CPs, and that Kir and KATP channels mediate this effect.
We thank the reviewer for the constructive feedback. We have revised our manuscript with respect to the referees’ comments. We have performed several additional experiments and analyses to address the concerns raised which helped to strengthen the conclusions of the study.
General comments/questions
Data are dumped wholesale in parenthetic expressions together with statistical information and figure call-outs, making it difficult to assess relationships. Such data-rich sentences should be re-written to more closely juxtapose data and associated conditions for more facile comparison. For example (lines 201-203):
Before: "Comparing the signal frequency between in vivo and ex vivo, there was a more than twofold reduction in calcium activity in both somata and processes (in vivo, S: 7.8 (2.5) and P: 18.9 (9.4); ex vivo, S: 3.1 (1.6) and P: 7.3 (5.6). Values are mean (SD) signals/min. Figure 3A)."
After: "A comparison of calcium activity showed that calcium signal frequency, expressed as mean ± standard deviation (SD), was reduced more than 2-fold ex vivo compared with that in vivo in both processes (7.3 ± 5.6 vs. 18.9 ± 9.4) and somata (3.1 ± 1.6 vs. 7.8 ± 2.5) (Figure 3A)."
Thank you for pointing this out. We have re-written data-rich sentences to facilitate the reading and comparisons of statistical information.
Distinct basal calcium transients of mural cells in vivo
Lines 126-7: "Basal calcium signals in SMCs and EPs 126 were usually only visible during the first 30 minutes of anesthesia." When exactly during this 30-min period were Ca2+ events in EPs recorded? Immediately after delivering anesthesia (i.e., before anesthesia effects on signaling manifested)? Some time during the 30-minute period? At the end of the 30-minute period? Measurements made immediately after delivering anesthesia, though more difficult to obtain, might provide physiologically meaningful information, but the only thing learned from responses measured during or at the end of the 30-minute period is how anesthesia affects Ca2+ signaling in EPs.
Ca2+ imaging data from SMCs and EPs were acquired within 25 min after the onset of isoflurane anesthesia. This is now clarified in the manuscript and in the method section.
Moreover, to rule out any possible anesthesia effects on the observed differences in Ca2+ dynamics between mural cells in vivo, we have now performed additional experiments in awake animals. The new data of awake imaging is now included in new Figure 2L-N. Importantly, although there were some differences in mural Ca2+ dynamics between anesthesia and awake imaging (see new Figure 2 —figure supplement 4), the differences between EPs and CPs regarding calcium signal frequency and duration were very consistent also in awake animals.
Persisting calcium signals in CPs ex vivo
Line 204-5: In many cases (Figure 3B), arteries and arterioles were either collapsed due to loss of tone or intraluminal pressure in the slice preparation or were constricted by possibly dead SMCs or EPs (no detectable calcium signal). These observations are disturbing signs of possible tissue injury and raise questions about the validity of data obtained using acute brain slice preparations. Faced with these observations, the authors should confirm cell viability in brain slice preparations after all pharmacological interventions that reduce Ca2+ signaling (e.g., hM3D(Gq)-DREADD/clozapine, SKF) to rule out the possibility that observed decreases in Ca2+ events reflect the trivial case of signal rundown in dying cells.
We agree with the reviewer, that assessment of cell viability in acute brain slices is crucial for proper data interpretation.
A loss of vascular tone by lack of blood pressure in the acute slice preparation is inevitable. However, in our study we aimed to complement our in vivo observations with acute slice experiments, which allows better access for pharmacological interventions e.g. to gain some mechanistic insights underlying the Ca2+ signals in pericytes.
We regret that our initial phrasing (“were constricted by possibly dead SMCs or EPs (no detectable calcium signal)”) was misleading. To better understand why SMCs and EPs showed strongly reduced Ca2+ signals in acute slice preparations, we have now performed additional experiments. Previous studies on SMCs and EPs in acute slice preparations were done in the presence of the widely used constricting agent U46619 (Mishra et al., 2014, Filosa et al., 2004). Strikingly, we revealed that initially “silent” SMCs and EPs regain Ca2+ activity upon bath-application of 100 nM U46619 (See new Figure 3E, F), thus showing that these cells are responsive and viable in our slice preparations. This also clarifies our previous observations of low to no Ca2+ signals in SMCs and EPs in the absence of vascular preconstruction (new Figure 3C, D). Incidentally, a recent study by the Nelson lab (Gonzales et al., 2020) confirms this finding in a retinal preparation.
Furthermore, to rule out that the reduction of Ca2+ signals observed in CPs by pharmacologic intervention with SKF (previous Figure 2E) is not simply a calcium run-down effect, we performed additional experiments with SKF and included an end-stimulus (addition of 100 nM U46619), to elicit a Ca2+ response (as shown in Figure 3D). We confirmed that SKF decreases Ca2+ activity in CPs, however, more importantly, this was not due to a run-down effect since CPs were viable and responsive, as shown by the large Ca2+ increase upon U46619 stimulation following SKF incubation (see new Figure 4E).
We would like to point out, that dying cells usually exhibit a sustained elevation of Ca2+ due to inactive ion-pumps and loss of membrane integrity (Carafoli and Krebs, 2016) and thus appear as very bright (Hill et al., 2017) and rounded cells, which we never observed during our experiments with pharmacological interventions. In the other cases where we observed a transient drop in Ca2+ activity, such as with sodium azide, hypoxia or potassium, Ca2+ signaling activity was always restored after washout, as shown in the traces (Figures 6A/D/E).
Furthermore, we want to clarify that all clozapine experiments were performed in vivo and not in acute slices. Nonetheless, we have now included new experiments to rule out possible clozapine induced side effects (see also further below):
1)We included Ca2+ recordings of CPs outside the virally transfected area of hM3D(Gq)-DREADD expressing neurons and these cells did not show any changes in Ca2+ signaling frequency upon clozapine treatment (see new Figure 5 —figure supplement 1).
2) Mural cells that showed a drop in Ca2+ signaling frequency in response to neuronal (DREADD) activation were monitored a day later and the same cells exhibited normal Ca2+ activity (new Figure 5 —figure supplement 2), clearly ruling out clozapine-induced cell death for the neuronal stimulation-evoked decrease in mural cell Ca2+ activity.
Line 226-7: "…massive calcium response in CPs when U46619 was included in the superfusate (Figure 3D, 226 Video 5). This overt calcium response in CPs was accompanied by cytoplasmic extrusions, suggesting that U46619 might be toxic for pericytes." Although U46610 can have harsh effects, including pericyte blebbing, the severity of U46610 effects described here suggest that procedures used to isolate, prepare and/or maintain brain slices increase the vulnerability of the preparation to potentially toxic stimuli.
We understand the reviewer’s concerns, however, we do not share the opinion that the U46619 effects observed were due to bad slice preparation or increased vulnerability of slices in our hands (see also previous comments regarding cell viability and patch clamp recordings of neurons further below).
We would like to point out that CPs were only stimulated by a short pulse of U46619. This led to a strong Ca2+ increase accompanied by changes in cellular morphology. However, since the cytosolic Ca2+ levels of the stimulated cells returned back to baseline after washout of U46619 (Figure 3D), we conclude that the cellular Ca2+ extrusion mechanisms were still intact. Further, the morphological changes of CPs were reversed after a prolonged washout of U46619(> 30 min) which we now added to (new Figure 3G, H).
Moreover, the morphological changes in U46619 stimulated CPs could hint to a cellular mechanism involving cytoskeletal rearrangements, as has been observed in an earlier study (Fernandez-Klett et al., 2010). Additionally, membranous blebs have been associated to cytoskeletal rearrangements (Robertson et al., 2021, Kondrychyn et al., 2020) resulting in changes in mechanical properties of endothelial cells.
Since U46619 has been shown to result in CP constriction (Fernandez-Klett et al., 2010), our observations on morphological changes of CPs suggest similar cytoskeletal rearrangements involved in movement/constriction. Moreover, also others recently reported the occurrence of membranous blebs along vasoconstriction in response to optogenetic stimulation of CPs (Hartmann et al., 2021).
We have discussed this possibility in the revised manuscript and added new figures to better showcase these transient morphological changes in CPs (see new Figure 3G, H and Figure 3 —figure supplement 2).
Line 233: Header for Figure 3 legend (Calcium dynamics in the absence of blood flow ex vivo is more affected in EPs than CPs) implies that differences in EP Ca2+ dynamics between the in vivo and ex vivo setting is caused by the absence of blood flow in the latter preparation. This is an overstatement; no data are provided to support a direct relationship between blood flow and EP Ca2+ dynamics.
We agree with the reviewer that this was an overstatement. We changed the header of Figure 3 as follows: “Mural cell calcium dynamics in acute cortical brain slices”.
Figure 3 (all): Are these all z-stacks/z-projections of vessels? If so, this should be indicated in the figure and/or figure legend. The appearance of pinching or narrowing could be due to vessel orientation relative to the imaging plane.
We thank the reviewer for pointing this out. Some images in Figure 3 were updated (also with the new data about SMC and EP Ca2+ activity). Images in Figure 3B and 3C are z projections and this has been now clarified in the figure legend.
CP calcium events are evoked by vasomodulators
Line 279: "Application of Nimodipine (100 μM)…" What is the rationale for using 100 μM nimodipine? This is an astonishingly high concentration for a dihydropyridine-even in a brain slice. A rationale is similarly required for SKF, which is also normally used at considerably lower concentrations than 100 μM.
The higher concentrations for these drugs were chosen to reach sufficient inhibition. Performing a dose-dependent inhibition analysis on CP Ca2+ dynamics with nimodipine and SKF would be out of the scope for a revision.
Lines 288-9: "We applied tetrodotoxin (TTX), a voltage-gated sodium channel blocker, to dampen neuronal activity (Zonta et al., 2003)….CP calcium signals were not affected" The absence of an effect of TTX could simply indicate that cells in the brain slice preparation are unresponsive because they are dead or dying. The viability of cells in the prep needs to be verified.
Ca2+ signals in CPs were persisting and were not affected by addition of TTX, therefore we conclude that the observed cells were viable. Of note, dead cells would not show spontaneous activity (Wu et al., 2019).
However, to showcase that also in our slice preparations cortical neurons are viable and fire action potentials, we now added new data (new Figure 6 —figure supplement 1). As expected, action potential firing in L2/3 neurons was completely abolished by TTX. Moreover, intrinsic membrane properties of these neurons were comparable to previous reports (Oswald and Reyes, 2008, Brown et al., 2019, Avermann et al., 2012).
Potassium released by neuronal stimulation leads to a calcium signal drop in CPs
Lines 322-4: "We used chemogenetics (Roth, 2016), in which neurons – in this case transduced to express hM3D(Gq)-DREADD – were activated with 30 μg/kg iv. clozapine." Transduction procedure needs to be described in Methods, and possible toxicity associated with it needs to be tested by assessing cell viability before and after transduction. In the absence of these latter control experiments, the possibility that the loss of Ca2+ signaling results from toxic effects of clozapine or toxicity of the transduction procedure itself cannot be ruled out.
We understand the concerns of the reviewer, that AAV transduction could be toxic. Our lab has longstanding experience in using AAVs for in vivo expression of genetically encoded sensors without inducing neuronal cell death, reactive astrogliosis and microglia activation (Zuend et al., 2020, Stobart et al., 2018, Machler et al., 2016). If transduction would be toxic, there should be no Ca2+ signals detectable in mural cells, which was not the case.
We described the transduction procedure in the Materials and methods under section “Virus injection and cranial window implantation:
(Lines 704-708: “A 4 x 4 mm craniotomy was performed above the somatosensory cortex using a dental drill (Bien-Air Dental), and for experiments requiring chemogenetics, adeno-associated virus (AAV) vectors were injected into the primary somatosensory cortex to achieve a localized chemogenetic receptor protein expression: 50 nl of AAV2-hSYN-hM3D (Gq)-mCherry (titer 1.02 x 1011 VG/ml; viral vector core facility (VVF), University of Zürich) at a cortical depth of 300 µm”).
As mentioned in an earlier comment (see above, point 3), we now included new experiments to rule out possible clozapine-induced side effects or toxicity.
1) We included Ca2+ recordings of CPs outside the virally transduced area of hM3D(Gq)-DREADD expressing neurons and these cells did not show changes in Ca2+ signaling frequency upon clozapine treatment (see new Figure 5 —figure supplement 1).
2) Mural cells that showed a drop in Ca2+ signaling frequency in response to neuronal (DREADD) activation were monitored a day later and the same cells exhibited normal Ca2+ activity (new Figure 5 – figure supplement 2) ruling out clozapine-induced cell death for the neuronal stimulation evoked decrease in mural cell activity.
Lines 332-338: Based on the set up for this section ("potassium is sensed by SMCs, causing the suppression of calcium oscillations"), the authors seem to be operating under the assumption that 10 mM K+ and BaCl2 are acting directly on mural cells-CPs in the current context. This is an incredibly bold assumption given that effects of 10 mM K+ and BaCl2 could be indirect through hyperpolarization of endothelial cells, which express Kir2.1 channels that are robustly activated by extracellular K+ and are blocked by 100 μM BaCl2.
Effects of KATP channel blockers could similarly be indirect through actions on endothelial cells. Although RNA seq data would suggest a large role for these channels in pericytes, further confirmation of their function should be provided, such as activation with pinacidil and block of the KATP channel during metabolic changes.
We agree with the reviewer that pharmacological experiments in acute brain slices are not celltype specific. Given the tight association of CPs with capillary endothelial cells there is a fundamental methodological problem to disentangle effects on either cell type. Indeed, both cell-types express the same channels that may influence cellular function in similar ways (e.g. see also a recent preprint of the Nelson lab reporting in retinal preparations functional expression of KATP channels in capillary endothelial cells and CPs (Sancho et al., 2021). Future studies employing more refined methods, such as cell-type specific knock-out animals or cell-type specific drug targeting could shed more light onto the intercellular relationship of CPs and capillary endothelial cells. However, at the moment there is no direct evidence of how changes in cortical endothelial cells impact CP behavior and vice versa.
Nonetheless, we are fully aware that we cannot base the effects of potassium, barium or KATP blockers solely on CPs, and we never intended to phrase it that way. We are aware of the effect of potassium and barium on endothelial cells and have acknowledged these studies several times in the manuscript. Whether the observed potassium induced Ca2+ signaling drop in CPs is a direct effect or via indirect actions involving capillary endothelial cells, cannot be properly disentangled at the moment.
To clarify these points, we have now changed the abstract, Results section and discussion and also included the possible cellular interaction between endothelial cells and CPs in our New summary Figure 7.
New discussion: “Potassium is released in high amounts during the repolarization phase after action potential firing (Paulson and Newman, 1987) and can act as a potent vasomodulator (McCarron and Halpern, 1990). Potassium sensing by SMCs and capillary ECs via Kir2 channels has been previously described as a mechanism to increase local cerebral blood flow (Longden and Nelson, 2015, Longden et al., 2017, Filosa et al., 2006, Haddy et al., 2006). Furthermore, a modeling study showcases the capillary EC Kir2 channel as a sensor of neuronal activity and highlights its impact on potassium – mediated neurovascular communication (Moshkforoush et al., 2020). […] Moreover, CPs are optimally positioned in the capillary bed (like cellular antennas of the EC-CP unit) to sense the microenvironment (Pfeiffer et al., 2021) and to amplify the hyperpolarization-mediated vascular response. Future studies using concurrent calcium imaging in ECs and CPs as well as cell-specific knock-out models are needed to gain more insights into the individual and the intercellular contribution of capillary ECs and CPs in potassium sensing and neurovascular coupling.”
Overall
Using predominantly an ex vivo acute brain slice preparation, the authors report differences in Ca2+ signaling signatures between EPs and CPs, lending additional experimental weight to the growing body of evidence that arteriole-proximate and more distal pericytes are functionally distinct. They describe differential effects of anesthesia on basal Ca2+ signaling in EPs (gradually eliminated) and CPs (largely unaffected) and show that Ca2+ signaling in EPs is reduced in brain slices. They further intimate (e.g., see Abstract, Figure 3 legend) that Ca2+ signaling in EPs requires blood flow; however, their data do not support this conclusion since flow effects on Ca2+ events were not tested directly. Importantly, the viability of cells in brain slices was not assessed following pharmacological manipulations that decreased Ca2+ signaling. In the absence of such control experiments, it is not possible to rule out tissue injury or improper slice maintenance as the cause of such decreases, especially given evidence for dead SMCs and EPs and unusually severe effects of U46619. Accordingly, some of the more compelling data from ex vivo pharmacological manipulations are suspect. The transient nature of EP Ca2+ responses recorded by 2-photon microscopy in cranial window model mice, reflecting signal dampening associated with the onset of anesthesia, similarly cast doubt on the significance of these in vivo findings. None of their data, including chemogenetic activation of hM3D(Gq)-DREADD-expressing neurons and pharmacological manipulations involving K+ and BaCl2, support the specific conclusion that neuronal K+ release causes the decrease/loss of pericyte calcium signaling. Importantly, pharmacological manipulations of Kir2 and KATP channels do not exclude endothelial cell contributions-especially given the importance of endothelial cell Kir2 in neurovascular coupling. More generally, the interconnected electrical properties of the microvasculature are not discussed or accounted for within the overall experimental design.
We appreciate the constructive feedback of the reviewer. We have now conducted a large battery of new experiments to address the concerns of (1) anesthesia by performing awake Ca2+ imaging and (2) cell viability (following manipulations that decreased Ca2+ signaling) by adding specific control measurements. We could also clarify the misleading notion of dead SMCs and EPs that appeared to be silent (low calcium activity) ex vivo in the absence of preconstruction. This resulted in several new figures which we presented and discussed specifically in the above responses. Moreover, we have discussed in the manuscript that we cannot exclude endothelial cell contributions and have now included an extended discussion on the interconnected electrical properties of the microvasculature, and the intriguing question whether pericytes could play a role in this complex cellular interplay.
Reviewer #2:
This is an exciting and timely study to uncover the mechanisms and physiological relevance of pericyte calcium signaling. By using pericyte specific calcium sensing-reporter mice (Pdgfrbeta-CreERT2; GCaMP6s (Ai96-flox)), Gluck et al. identify distinct Ca signatures within the heterogeneous populations of pericytes along the brain vasculature including smooth muscle cells (SMCs), ensheathing pericytes (EPs; found along the branch orders 1-4), and capillary pericytes (CPs; found along > 4th order). While SMCs and EPs have synchronous calcium oscillations that are expectedly related to vessel diameter changes, CPs have asynchronous calcium signaling in somata and processes and there is no correlation with vessel diameter changes in the healthy anesthetized animal. This has been shown in some prior studies, but the current manuscript delves far deeper into the basis of pericyte calcium signals.
To pharmacologically test how calcium signals are regulated in pericytes, they moved to ex vivo, brain slice experiments. They show that blocking VGCC and TRPC reduces CP calcium signals. Activating neurons using chemogenetics reduced CP calcium signals in vivo. KCl, to mimic neuronal activity, also reduced CP calcium and this effect is attenuated by blocking Kir2 and Kir6.1 ex vivo. Creating hypoxia-like conditions (in vivo: N2 or ex vivo: sodium azide) reduced calcium signals in CPs. Conversely, a number of vasoactive substances known to constrict arterioles lead to large calcium influxes in CPs, including ET-1, ATP and UDP-glucose.
These studies pave new roads to understanding how pericytes are involved in NV coupling, and blood flow control. The approach is highly innovative and the in vivo-ex vivo approach is a major strength. However, there is a missing link that makes it difficult to understand the physiological relevance of the CP calcium signals as currently presented (comments 1 and 2). If addressed, this would substantially increase the impact of the work.
1) It is not clear how calcium signaling in CPs is related to electrical conductance along the capillaries leading to arteriole diameter change. For example, how do CP calcium changes relate to calcium in SMCs and EPs (and/or arteriolar diameter) on the same vascular tree during NV coupling in vivo? How does CP calcium levels relate to endothelial activity? Is it possible to block CP K+ signaling in vivo to test the hypothesis that upstream arteriolar dilation will be decreased? Perhaps the relatively selective expression of KATP in CPs can be used to leverage an in vivo pharmacological experiment. It is not necessary to address all these questions, but some additional data to link CP calcium to arteriolar dilation would strengthen the paper and substantiate the compelling model in Figure 6.
We thank the reviewer for his positive and constructive feedback. We fully share the reviewer’s curiosity and are highly interested in how CP Ca2+ signals relate to NVC. Despite the complexity of the question, we performed additional experiments in vivo with an intent to address how neuronal activity-evoked changes in CP Ca2+ dynamics associate with SMCs and EPs and arteriolar diameters along the same vascular tree. We limited the neuronal DREADD (Gq) expression by AAV delivery close to the capillary bed, however, we cannot completely exclude partial neuronal expression in the vicinity of arterioles or arteries. When we activated neurons with clozapine, we observed a drop in Ca2+ activity in all mural cells along a connected vascular tree (see new Figure 5). This drop in Ca2+ activity was paralleled by a vessel diameter increase in all vascular compartments, i.e. artery, arteriole and capillary. Thus, we can say that the drop in CP Ca2+ activity associates with an increase in capillary diameter, however, it remains open whether this capillary diameter change is mediated by CPs or is an indirect effect of upstream artery/arteriole diameter changes. For now, we cannot answer how this relates to possible changes in endothelial activity (this would require e.g. simultaneous dual Ca2+ imaging in CPs and endothelial cells) and further investigations with different approaches are needed to address these questions. These could include sparse optogenetic stimulation of neurons close to CPs, and cell-specific conditional mutants (targeting e.g. Kir or KATP channels, and/or gap junctions) where the interaction between CPs or between CPs and endothelial cells could be studied. In light of recent findings showing how optogenetic activation of CPs may lead to capillary constriction, as shown by the reviewer’s lab (Hartmann et al., 2021), yet sensory stimulation did not evoke measurable capillary diameter changes (Hill et al., 2015) it may well be that a drop in CP Ca2+ activity (in response to certain levels of neuronal activity and increases in extracellular potassium) could be involved in facilitating capillary dilation. However, future studies are needed to fully resolve the role of CPs in neurovascular coupling. We have re-written the results and discussion to clarify these points and point out unresolved questions and future directions.
To further investigate mechanisms of signal transmission between CPs, we imagine an experiment (Author response image 1), under the hypothesis, that neuronal activity leads to a Ca2+ drop in CP (CP1) and that this Ca2+ drop is transmitted upstream along connected CPs (CP2, CP3) on a vascular branch. Using laser-ablation of a CP (CP2) that is in between connected CPs should possibly stop transmission of a Ca2+ drop. Further, the impact of gap-junction coupling between endothelial cells or pericytes and endothelial cells could be investigated by pharmacologically blocking gap-junction proteins or by cell-type specific knockouts.

Hypothetical experiment to investigate signal transmission between CPs along a vascular branch.
Furthermore, it is possible to perform dual Ca2+ imaging of endothelial cells and pericytes by either crossing acta2-RCaMP1.07 (JAX Stock.: 028345) and Cdh5BAC-GCaMP8 (JAX Stock.: 033342) or our here used cross between Pdgfrb-CreERT2 (JAX Stock.: 029684) and GCaMP6s (JAX Stock.: 028866) in combination with an endothelial specific AAV (Korbelin et al., 2016) for RCaMP1.07 sensor expression in capillary endothelial cells.
Ultimately, electrophysiological recordings of pericytes and endothelial cells, combined with selective pharmacology could provide more specific insights into the nature of signal spread (as for example showcased in a preprint by the group of Mark Nelson (Sancho et al., 2021)).
2) Local regulation of capillary flow: The presumption is that the sustained calcium changes evoked by vasomediators (ET-1, ATP) and neural activity are affecting local capillary diameter. This would be helpful to know because past studies have not been as rigorous in reporting vascular branch orders, leaving some ambiguity as the authors note. The data would help us understand whether CPs are autonomously able to regulate capillary diameter. This is important even if the stimulation becomes supra-physiological, as it clarifies whether CPs are a logical target for constriction in pathologies such as stroke, AD. Curiously, some of the data also show a slower component of calcium modulation can occur on the scale of tens of second to minutes (traces of Figure 5). Are these slow modulations related to capillary diameter? Similarly, how does N2-induced hypoxia result in vascular diameter changes along the various pericyte territories in relation to Ca signal changes?
We thank the reviewer for these insights.
Indeed, we could observe capillary diameter changes (constrictions) when brain slices were treated with vasomediators ET-1 or U46619 which evoked a large Ca2+ increase in CPs (Figure 3D and Figure 3 —figure supplement 2). However, we are reluctant to provide diameter measurements for these observations made in brain slices, since we only detected these vasculature changes through staining of stationary blood plasma, without actually seeing the vessel borders. To be more conclusive, repeating these experiments in future studies with a dual imaging approach using epifluorescent/transmitted light microscopy would yield more precise measurements of vessel diameters.
On the other hand, by increasing neuronal activity (Gq-DREADD activation) or transient hypoxia we could observe a drop in Ca2+ activity in CPs which associated with capillary diameter changes (dilation) in vivo (new Figure 5D, new Figure 6 —figure supplement 3). These measurements were acquired from timeaveraged (20 s) images of the vasculature.
Taking these two observations together, it is plausible that silencing and increasing CP Ca2+ activity could be involved in regulating capillary diameters. Our findings are in line with the recent report of how optogenetic stimulation of CPs induce capillary constrictions (Hartmann et al., 2021). How Ca2+ changes and cytoskeletal rearrangement in CPs regulate capillary diameter, and whether this occurs at the level of individual CPs or involves intercellular crosstalk with other mural cells and endothelial cells, needs to be determined in future studies.
3) Neural activity: DREADDs are gated by clozapine-N-oxide, rather than clozapine. It is correct that clozapine was used in this study, instead of its less bioactive counterpart? If so, a control non-DREADD expressing group might be needed to verify the intended goal of stimulating local neurons in vivo, as opposed to broad actions of clozapine. Further, it is suggested that basal neuronal activity in slice does not regulate CP calcium fluctuations because TTX has no effect. However, neuronal activity was not actually measured.
We thank the reviewer for his comment. We have chosen clozapine over clozapine-N-oxide (CNO) based on following reasons:
1) It has been shown, that clozapine is actually the effective molecule in DREADD activation, since CNO is back-converted to clozapine in vivo (Gomez et al., 2017).
2) Due to back-conversion of CNO to clozapine, CNO is in fact less bioactive than clozapine, requiring higher doses, which may in turn lead to unwanted side-effects (Jendryka et al., 2019).
We used clozapine at a dose of 30 µg/kg which is in the previously reported concentration range void of side-effects (Cho et al., 2020).
As discussed in an earlier response to reviewer 1 (see above, point 3 and 9), we showed that clozapine treatment itself had no direct effect on Ca2+ signaling in CPs when measured in an area outside of hSynGq-DREADD transduced neurons (new Figure 5 —figure supplement 1). Furthermore, we also show that the neuronal activation induced drop in Ca2+ in mural cells was not due to any lasting toxic side effects, since the same mural cells revealed normal Ca2+ activity when imaged on the following day (new Figure 5 —figure supplement 2).
Our findings of neuronal activity induced decrease of Ca2+ activity in cortical CPs are in line with a recent study showing how neuronal stimulation with odorants reduces CP Ca2+ levels in the olfactory bulb (Rungta et al., 2018).
It is indeed very intriguing that an increase in neuronal activity impacts CP Ca2+ dynamics. It could be that CP Ca2+ levels are regulated primarily by an acute rise in neuronal network activity (highly raising extracellular potassium concentrations). This could possibly explain why sensory whisker stimulation was reported to not influence CP Ca2+ dynamics (Hill et al., 2015).
Indeed, our TTX slice experiments revealed that basal neuronal activity (which is likely low in slices) has no major influence on CP Ca2+ dynamics (see Figure 6 —figure supplement 2). We have not measured neuronal activity during CP Ca2+ imaging, but we provide additional data showing that TTX silences action potential firing in our slice preparations (new Figure 6 —figure supplement 1). This may substantiate the notion that CPs primarily react (by decreasing Ca2+ activity) to acute changes of elevated neuronal activity. However, to define a specific level or type of neuronal activation requires future investigation.
4) In addition to a Non-DREADD expressing control, some additional controls would increase rigor. To ensure that the fluctuation in calcium dynamics in pericytes is not influenced by animal motion or passing RBCs in vivo, comparison to a general fluorescent reporter like GFP would be helpful. Vehicle control experiments are not described for the ex vivo pharmacological experiments. If these were performed, it would be good to show.
We agree with the reviewer, that control experiments are important to increase rigor.
As mentioned above, mural cell Ca2+ imaging was performed in a non-DREADD expressing brain region as internal control (new Figure 5 —figure supplement 1).
We have performed vehicle control experiments which are now added as a supplementary figure (new Figure 4 —figure supplement 1).
We are confident that the fluctuations in pericyte Ca2+ dynamics in vivo were not influenced by passing RBCs or motion given that CP Ca2+ signaling frequency were comparable to the recordings in acute brain slice conditions without any influence of passing RBCs and motion.
Reviewer #3:
The Manuscript by Glück et al. examined the patterns of calcium transients in vascular mural cells in the live mouse brain and brain slices using two photon imaging of calcium sensor mice, in combination with pharmacological and chemogenetic manipulations. They found that pericytes exhibit distinct calcium dynamics compared to vascular smooth muscle cells (vSMCs). Pericytes in vivo have compartmentalized, irregular, higher-frequency calcium transients, which are not correlated with vessel diameter changes, whereas calcium fluctuation of vSMCs are highly correlated with changes in vessel diameter. In ex vivo preparations, pericytes retained their spontaneous calcium properties in contrast to its loss in SMCs. Combined with chemogenetics in vivo and pharmacological manipulation in slices, the authors demonstrated neuronal activity can decrease calcium response in pericytes by elevating extracellular potassium, which is mediated by Kir2.2, Kir6.1. Energetics state, and hypoxia challenge can also affect pericyte Ca2+ activation through KATP channel.
Overall, this paper is technically rigorous and reproduces the majority of in vivo findings by recent publications (PMIDs: 26119027, 29937277). It also provides a more granular investigation of Ca2+ transients in pericytes (comparing soma and processes) and the pharmacological investigation of potassium channels and ATP in slices is also novel.
We thank the reviewer for his positive feedback to our work.
1) The main criticism we have about this paper is the nomenclature used to define the different mural cells which continues to confuse the field. Why use the title "signatures of calcium in brain pericytes" instead of simply saying in mural cells? This is surprising given that their data completely parallels data showing that SMCs and pericytes are distinct cells types. They introduce again the confusing term "ensheathing pericytes" despite their finding that EP essentially have identically characteristics to vSMCs (near identical calcium properties, contractility, expression of aSMA, circumferential anatomy, lack of compartmentalized Ca2+ signals and others) as well as findings from other groups including transcriptome data showing no such thing as subpopulations of pericytes (PMIDs: 29443965, 30129931) and data from uptake of small fluorescent molecules specifically by CP but not other mural cells (PMID: 28504673). We think it would be critical for this paper to clarify the field rather than continue to confuse the nomenclature and cite historical 100 old anatomical studies to justify it. Otherwise it is essential that they produce functional, structural and genetic evidence that there is such thing as a distinct ensheathing cell that is closer to pericytes than to vSMCs to justify the name. The logical terminology would be to call them "terminal vSMCs" or something similar to account for the minor morphological difference these cells have compared to the slightly more proximal vSMCs.
We thank the reviewer for this critical point and adapted the title of the work to refer to mural cells. We now also included data of spontaneous Ca2+ signals in SMCs (new Figure 2).
Regarding the term ensheathing pericyte (EP), we see the issue as being whether to name a cell after a function (vasomotor activity by smooth muscle actin) or morphology and earliest description of pericytes (Zimmermann, 1923). The apparently continuous transition in morphology of mural cells from SMCs to pericytes and a missing consensus on how to label different vascular segments further complicates the issue (Holm et al., 2018). In recent literature, vascular branches following an artery were called arteriole, precapillary arteriole or directly capillary. Classification of mural cell subpopulations along the vasculature according to branch orders (Grant et al., 2017) could help to make future studies more comparable between each other. We therefore chose to adapt this branch ordering system in our study as well. Although transcriptomic studies (Vanlandewijck et al., 2018, He et al., 2018) of mural cells suggest no subpopulation of pericytes, three different SMC clusters (arterial (a), arteriole (aa) and venular (v) SMCs) were found. However, the differentiation between aaSMCs and pericytes was based on the expression of smooth muscle actin (Acta2) which has been shown to stop abruptly at the transition from arteriole to capillary level. This approach directly defines cells on arterioles to be SMCs. Despite the fact that aaSMCs were found molecularly to be more closely related to aSMCs compared to capillary cells, these transcriptomic studies were not set to reveal more subtle changes between these cell clusters. Even more so, considering a relatively low abundance of EPs/aaSMCs compared to aSMCs or CPs in the reported dataset.
In terms of functional differences, a recent study by the Nelson lab found that these perivascular cells in question lack functional ryanodine receptors (RyR), which is a defining feature of smooth muscle cells (Gonzales et al., 2020).
Additionally, these perivascular cells (i.e. EP) indeed have a pericyte appearance with a protruding cell body and an approximately two-fold greater length along the longitudinal vascular axis (Berthiaume et al., 2021). We also think that the morphology speaks more to the site of pericytes. SMCs are ordered “rings” along the vessel with “sausage-like” nuclei (Author response image 2), while EPs show a protruding cell body with ovoid nuclei (Author response image 2ii), which is also the case with CPs (Author response image 2iii).

(i) Confocal image of SMCs showing Merge, GCaMP6s (pdgfrβ-driven), and nuclear DAPI stain.
(ii) Confocal image of an EP/terminal vSMCs showing Merge, GCaMP6s (pdgfrβ-driven), and nuclear DAPI stain. (iii) Confocal image of a CP showing Merge, GCaMP6s (pdgfrβdriven), and nuclear DAPI stain. Yellow arrows point to the respective cell-type. Scale bars: 10 µm.
Considering all of the above points, in our opinion, the jury is still out there whether to name the perivascular cell in question ensheathing pericyte or terminal vascular smooth muscle cell (or precapillary SMC). Therefore, we will state in the revised manuscript, that the term ensheathing pericyte (EP) is synonymously used with terminal vascular smooth muscle cell (vSMC). However, we are very open to discuss this further.
2) Although not essential for this project, it would have been nice to obtain concurrent endothelial and pericyte Ca recordings (using RCaMP and GCaMP sensors) to really understand the role of CP Ca2+. Is it similar to the described hyperpolarization propagation in endothelium? (PMID: 28319610), are there any interactions between CP and endothelium?
We agree with the reviewer that concurrent endothelial and pericyte Ca2+ recordings would be very helpful to understand the relationship between endothelial cells and CPs. However, this is beyond the scope of this work presented here and is an exciting question for a future study.
Several new studies hint to the importance of endothelial calcium signaling for adaptations of blood flow, shear forces and implications in pathologies, such as diabetes and hypertension (Thakore et al., 2021, Fancher and Levitan, 2020, Yamamoto et al., 2006).
Moreover, studies in retinal preparations could show an electrical signal transmission between adjacent CPs (Wu et al., 2006). Given the tight interactions of CPs and endothelial cells via peg-socket contacts (Ornelas et al., 2021) and possibly via gap junctions (Perrot et al., 2020), a similar hyperpolarization as in endothelial cells could also occur in CPs. Yet, further studies are needed to gain more insights into the signaling spread along these mural cells.
To address these open questions in the field we have added to the discussion the following sentences:
“However, Kir2 and KATP channels are also expressed in capillary ECs which mediate a hyperpolarization in response to increases in extracellular potassium (Longden et al., 2017). Given that there is evidence of EC and CP gapjunctional coupling in the retinal vasculature (Wu et al., 2006, Kovacs-Oller et al., 2020, Ivanova et al., 2017, Ivanova et al., 2019) and the lack of cell-type specificity of pharmacological manipulations, we cannot exclude that the observed potassium evoked calcium drop in CPs could be secondary to changes in capillary ECs. However, still very little is known of how capillary ECs interact and modulate CP functions or vice versa. Gap-junction coupling of capillary ECs and CPs would allow for electrical interconnection between these cells, forming a vascular relay (Ivanova et al., 2019). Activation of Kir2 and KATP channels likely induces cellular hyperpolarization in this EC-CP capillary unit, which could be transmitted between and within CPs and ECs through gap junctional coupling (Figure 7). Potassium sensing by CPs and ECs could potentiate the propagation of hyperpolarization from a site of elevated neuronal activity to upstream feeding arterioles.“
3) They should also comment on the fact that it is very difficult to target pericytes without having impact on other cell types by pharmacological manipulation. Only Kir6.6 (Kcnj8) blocking provides pericyte specificity, since BaCl2 can also affect endothelial Kir channels (PMID: 26840527), which may indirectly change pericytes calcium.
We agree with the reviewer and we have discussed in the revised manuscript the difficulty to interpret pharmacological manipulations regarding cell specificity (see also our response to comments of reviewer 1, point 10). We added several sentences (see also in the comment before) which discuss this issue.
4) Is there any directionality in the propagation of the Ca transients in pericytes in vivo?
This is a very interesting question. The way we recorded and quantified Ca2+ signals did not allow us to measure signal propagation. However, our impression is that there was no apparent signal directionality, yet this requires a more detailed analysis and employment of fast volumetric scanning of whole pericytes.
5) Do the microdomains of Ca in vitro become more synchronized between processes and Soma. If so, could this mean that in the in vitro prep the microenvironment around pericytes is more homogeneous than in vivo, thereby causing more homogeneity of Ca?
This is an interesting insight. We performed a cross correlation analysis to assess, whether process transients were co-occurring with somata transients. Comparing awake, anesthetized and ex vivo measurements revealed that process transients in awake mice were occurring on average 2 s before a soma transient (linear mixed model, post-hoc p = 0.025). However, this effect is skewed, since 99.5 % of all process transient peaks are in the same range (-10 to 10 s time to Soma (see histogram, middle panel of Author response image 3), grey dots represent outliers) as anesthetized and ex vivo measurements. There were no significant differences between anesthetized and ex vivo measurements (Author response image 3).

Correlation analysis of process and somata transients for ex vivo, anesthetized and awake Ca2+ – measurements.
Grey points represent outlier values.
6) In their figure 6 model there is no mention of possible pericyte to pericyte or pericyte to SMC communication. This would be worth considering as these cells are likely to be gap junction coupled.
We thank the reviewer for pointing out this possibility. We updated the previous figure 6 (now new Figure 7) to include the possibility of pericyte to pericyte communication. See also our response to the last remark of Reviewer #1.
https://doi.org/10.7554/eLife.70591.sa2Article and author information
Author details
Funding
National Science Foundation (310030_182703)
- Chaim Glück
- Annika Keller
- Jillian L Stobart
- Bruno Weber
National Science Foundation (31003A_156965)
- Chaim Glück
- Annika Keller
- Jillian L Stobart
- Bruno Weber
The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We thank the Viral Vector Facility of the University of Zürich for the supply of AAV vectors. We are grateful to Marc Zünd for assembly and maintenance of two-photon microscopes. Furthermore, we thank Zoe Looser for genotyping help. Karen Everett is thanked for critical proofreading of the manuscript. This project was supported by the Swiss National Science Foundation (Grant 310030_182703 and 31,003A_156965).
Ethics
All animal experiments were approved by the local Cantonal Veterinary Office in Zürich (license ZH 169/17) and conformed to the guidelines of the Swiss Animal Protection Law, Swiss Veterinary Office, Canton of Zürich (Animal Welfare Act of 16 December 2005 and Animal Protection Ordinance of 23 April 2008). Every effort was made to minimize suffering and conform to the 3Rs principles.
Senior Editor
- Kenton J Swartz, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
Reviewing Editor
- Mark T Nelson, University of Vermont, United States
Reviewers
- Mark T Nelson, University of Vermont, United States
- Jaime Grutzendler, Yale University, United States
- Andy Shih
Publication history
- Received: May 21, 2021
- Accepted: June 15, 2021
- Accepted Manuscript published: July 6, 2021 (version 1)
- Accepted Manuscript updated: July 16, 2021 (version 2)
- Version of Record published: July 21, 2021 (version 3)
Copyright
© 2021, Glück et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 2,131
- Page views
-
- 295
- Downloads
-
- 15
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Medicine
- Neuroscience
The cell bodies of postganglionic sympathetic neurons innervating the heart primarily reside in the stellate ganglion (SG), alongside neurons innervating other organs and tissues. Whether cardiac-innervating stellate ganglionic neurons (SGNs) exhibit diversity and distinction from those innervating other tissues is not known. To identify and resolve the transcriptomic profiles of SGNs innervating the heart, we leveraged retrograde tracing techniques using adeno-associated virus (AAV) expressing fluorescent proteins (GFP or Td-tomato) with single cell RNA sequencing. We investigated electrophysiologic, morphologic, and physiologic roles for subsets of cardiac-specific neurons and found that three of five adrenergic SGN subtypes innervate the heart. These three subtypes stratify into two subpopulations; high (NA1a) and low (NA1b and NA1c) neuropeptide-Y (NPY) -expressing cells, exhibit distinct morphological, neurochemical, and electrophysiologic characteristics. In physiologic studies in transgenic mouse models modulating NPY signaling, we identified differential control of cardiac responses by these two subpopulations to high and low stress states. These findings provide novel insights into the unique properties of neurons responsible for cardiac sympathetic regulation, with implications for novel strategies to target specific neuronal subtypes for sympathetic blockade in cardiac disease.
-
- Neuroscience
Oscillations of extracellular voltage, reflecting synchronous, rhythmic activity in large populations of neurons, are a ubiquitous feature in the mammalian brain, and are thought to subserve important, if not fully understood roles in normal and abnormal brain function. Oscillations at different frequency bands are hallmarks of specific brain and behavioral states. At the higher end of the spectrum, 150-200 Hz ripples occur in the hippocampus during slow-wave sleep, and ultrafast (400-600 Hz) oscillations arise in the somatosensory cortices of humans and several other mammalian species in response to peripheral nerve stimulation or punctate sensory stimuli. Here we report that brief optogenetic activation of thalamocortical axons, in brain slices from mouse somatosensory (barrel) cortex, elicited in the thalamorecipient layer local field potential (LFP) oscillations which we dubbed “ripplets”. Ripplets originated in the postsynaptic cortical network and consisted of a precisely repeating sequence of 2‑5 negative transients, closely resembling hippocampal ripples but, at ~400 Hz, over twice as fast. Fast-spiking (FS) inhibitory interneurons fired highly synchronous 400 Hz spike bursts entrained to the LFP oscillation, while regular-spiking (RS), excitatory neurons typically fired only 1-2 spikes per ripplet, in antiphase to FS spikes, and received synchronous sequences of alternating excitatory and inhibitory inputs. We suggest that ripplets are an intrinsically generated cortical response to a strong, synchronous thalamocortical volley, and could provide increased bandwidth for encoding and transmitting sensory information. Importantly, optogenetically induced ripplets are a uniquely accessible model system for studying synaptic mechanisms of fast and ultrafast cortical and hippocampal oscillations.