Introduction

Astrocytes play important roles in the regulation of BOLD signals through the neurovascular coupling^1–3. Recently, Wang et al.⁴ reported that sensory stimulation evoked astrocytic calcium signals were correlated with positive BOLD signals, while state-specific astrocytic calcium signals were correlated with negative BOLD signal in anesthetized rats. However, the causal evidence regarding the contribution of astrocytic calcium activity to the BOLD signal is lacking. In this study, we combined (1) awake mice fMRI with neuronal and astrocytic calcium recording, (2) astrocytic manipulation and (3) video monitoring of behavioral states to elucidate the mechanism of the bidirectional contributions on BOLD responses from stimulus-evoked and state-specific astrocytic calcium signals.

Methods

Seven mice were used in each awake mouse fMRI experiment. The viruses used for labeling S1BF neuronal, astrocytic signal or activation/inhibition of the S1BF astrocytes were shown in Main Figures (Fig.1-4), respectively. Optical fiber and head holder implantation were performed one week after virus injection. The spectrally resolved fiber photometry system was shown in Fig. 1, which was used for recording fluorescence signals during the fMRI scanning.
All MRI data were collected using multiband EPI sequence with following parameters: TR=750 ms, TE=15 ms, flip angle=46.8°, bandwidth=300 kHz, field of view=15×10.5 mm², matrix size=100×70, slice thickness=0.48 mm, 1200 volumes per scan, and 38 axial slices per volume. A T2-weighted structural image was acquired for co-registration with following parameters: TR=3400 ms, TE=33 ms, field of view=16×16 mm², matrix size=256×256, slice thickness=0.5 mm, 30 axial slices, RARE factor=8, and number of averages=2. Each simultaneous fiber photometry and fMRI session included 5 or 8 EPI scans with whisker stimulations and the same number of EPI scans in resting state. In addition, two custom-made MR-compatible video cameras (25 fps) was placed inside the bore to record the mouse spontaneous behaviors.
After basic preprocessing procedures of fMRI data, including rigid-body and nonlinear spatial coregistration and 0.4-mm spatial smoothing, the BOLD signals were further regressed by “6 rp+6 Δrp+10 PCs” nuisance signals to minimize the effects of scanner drift, motion and other non-neural physiological noises adopted from our previous study⁵, in which “6 rp+6 Δrp” nuisance signals represented 6 head motion parameters and their 1^st order first derivatives, and “10 PCs” were the first 10 principal components from the BOLD signals of non-brain tissue, e.g., the muscles.

Results

Fig. 1 showed the timeline of simultaneous fiber photometry and fMRI in awake mice. Gradient air pressure test showed that 16-psi whisker stimulation was sufficient to evoke robust S1BF neural and astrocytic activation (Fig. 1D-E). Combined with spontaneous behavioral monitoring by MR-compatible cameras during fMRI scanning (Fig. 2), we found whisker stimulation accompanied with body movement caused larger astrocytic calcium signal compared with whisker stimulation only (Fig. 2D-E). Such larger astrocytic calcium signal coupled to delayed negative BOLD response (Fig. 2E-F).
Thus, we hypothesized that stimulation evoked astrocytic calcium signals contribute to positive BOLD response, while spontaneous state-specific astrocytic calcium signals contribute to negative BOLD response (Fig. 3).
To validate our hypotheses, we firstly examined how the calcium elevation in S1BF astrocytes contributes to BOLD signals by elevating astrocyte calcium with a chemogenetic method, i.e., GfaABC1D-hM3Dq + clozapine (Fig. 4). We found whisker stimulation evoked higher astrocytic and positive BOLD responses in the task-aligned condition. However, in the concurrent (task-aligned and spontaneous) condition, whisker stimulation evoked more apparent increases of ipsilateral neuronal, astrocytic responses by chemogenetic activation of S1BF astrocytes, but no significant increases of ipsilateral positive BOLD signals (Fig. 4C). In the resting state, higher spontaneous astrocytic signal fluctuations induced higher ipsilateral neuronal responses but lower ipsilateral BOLD signal dynamics for both positive and delayed negative ones (Fig. 4D).
Moreover, we performed a loss-of-function experiment by selectively suppressing calcium elevation in S1BF astrocytes using the IP3R2^flox mice. Injection of GfaABC1D-Cre into the S1BF of the IP3R2^flox mice suppressed astrocyte calcium transients and then we found whisker stimulation evoked lower astrocytic and BOLD responses in both task-aligned only and concurrent conditions (Fig. 5B). In the resting state, inhibition of S1BF astrocytes reduced the amplitude of ipsilateral spontaneous BOLD signal dynamics (Fig. 5C).

Conclusion

In this study, we revealed the coupling between evoked and state-dependent astrocyte calcium signal to positive and negative BOLD signal respectively, and provided causal evidences through the activation and inhibition of S1BF astrocytic signals. These results demonstrated that evoked astrocyte calcium mainly contributes to positive BOLD signal and state-specific astrocyte calcium contributes to negative BOLD signal. Our results made great sense on understanding astrocytes serve distinct roles in directly sensing neuronal activity and feedback to arousal transients through different neurovascular coupling process.

Acknowledgements

This work was supported by the National Science and Technology Innovation 2030 Major Program (2021ZD0200100 to ZL), Strategic Priority Research Program of Chinese Academy of Sciences (XDBS01030100 to ZL), Pioneer Hundreds of Talents Program from the Chinese Academy of Sciences (to ZL), Shanghai Municipal Science and Technology Major Project (2018SHZDZX05 to ZL), the National Natural Science Foundation of China (82171899 to ZL), Lingang Laboratory (LG202104-02-06 to ZL), Postdoctoral Innovative Talents Support Program (BX20230383 to CT).