Rapid, biochemical tagging of cellular activity history in vivo

The study addresses the need for noninvasive, rapid, and stable tagging of cells experiencing elevated intracellular Ca2+ levels, particularly in deep tissues where optical access is limited. Existing genetically encoded Ca2+ indicators provide transient fluorescent signals and often require invasive optical access. Light-gated Ca2+-dependent integrators (e.g., FLARES, FLICRE, Cal-Light, CaMPARI) demand blue/UV light, limiting scalability in deep brain regions. Immediate-early-gene (IEG)-based transcriptional reporters (e.g., TRAP2, tetTag) rely on slower transcriptional responses and are not as universally coupled to activity as Ca2+. The authors hypothesize that an enzyme reconstituted by Ca2+-dependent protein–protein interaction can provide a rapid, reversible, and time-gated biochemical tag (biotin) to record recent cellular activity without light, enabling immediate post hoc readout and application in freely moving animals.

- Fluorescent Ca2+ indicators have transformed neural recording but are transient and require invasive optics for deep structures. - Light-gated Ca2+-dependent reporters (FLARES, FLICRE, Cal-Light) and CaMPARI allow integrative labeling but need blue/UV light, restricting use in deep tissues and in untethered animals. - IEG-based tagging systems (TRAP2, tetTag) use drug-gated windows but have slow onset (hours) and variable coupling to activity, complicating immediate tagging and detection. - Proximity-labeling enzymes (BioID, TurboID, split-TurboID) biotinylate proximal proteins over long timeframes and have not been adapted to detect dynamic ionic changes. The authors leverage split-TurboID to engineer a Ca2+-dependent, rapid activity tag.

Design and optimization of CaST: - Engineered Ca2+-activated split-TurboID (CaST) by fusing calmodulin (CaM) to one split-TurboID fragment (sTb(N)) and a CaM-binding peptide (M13 variant) to the other fragment (sTb(C)). Elevated Ca2+ recruits CaM to M13, reconstituting TurboID to biotinylate nearby proteins only when exogenous biotin is present. - Multiple configurations tested in HEK293T cells. Optimal architecture: membrane-tethered CD4-sTb(C)-M13-GFP and cytosolic CaM-V5-sTb(N). Best transfection ratio was 5:2 (CD4-sTb(C)-M13-GFP : CaM-V5-sTb(N)). - Implemented bicistronic delivery (P2A or IRES) to ensure coexpression. The IRES construct (CaST-IRES) provided higher Ca2+-dependent signal-to-background ratio (SBR) versus P2A and was used for subsequent characterization. HEK293T characterization: - Immunofluorescence with SA-647 detected biotinylated proteins; GFP reported expression of the CD4-sTb(C)-M13-GFP component; anti-V5 labeled CaM-V5-sTb(N). Western blots with SA-HRP confirmed biotinylation only with biotin + Ca2+. - Single-cell quantification across multiple fields of view (FOVs) computed SA-647/GFP ratios to normalize expression variability. - Reversibility test: Ca2+ (30 min), wash for 10 min, then biotin (30 min) produced no labeling vs. simultaneous biotin + Ca2+, indicating reversible reconstitution and ~10-min temporal resolution. - Ca2+ titration: treated with 0–10 mM CaCl2 plus 1 µM ionomycin and 50 µM biotin for 30 min; measured SA-647/GFP across FOVs. - Time integration: applied biotin + Ca2+ for 10–240 min to assess labeling kinetics and saturation. Comparison to FLICRE: - Transfected HEK293T cells with CaST-IRES or FLICRE. Applied 15-min stimulation (biotin + Ca2+ for CaST; blue light + Ca2+ for FLICRE). Fixed cells immediately or after 2, 4, 6, 8 h. Quantified normalized reporter readouts (SA-647/GFP for CaST; UAS::mCherry/GFP for FLICRE). Also tested higher expression levels (48 h post-transfection) to assess nonspecific background susceptibility. Neuronal culture experiments: - Rat hippocampal neurons infected with AAV2/1-Synapsin-CD4-sTb(C)-M13-GFP and AAV2/1-Synapsin-CaM-sTb(N). Stimulation paradigms: 30 mM KCl ± biotin for 10 or 30 min; pharmacology with 10 µM dopamine (DA), 10 µM DOI (5-HT2A/C agonist) ± biotin for 30 min. - RCaMP2 co-expression to correlate Ca2+ dynamics (dF/F peaks) with CaST SA-647/GFP on a per-neuron basis under mild stimulation (media change) + biotin for 30 min. - Spatial specificity: Co-expressed red-shifted opsin bReaChES; delivered orange light through a ~1-mm slit with synaptic blockers (APV 50 µM, NBQX 20 µM) during 30-min biotin exposure to test spatially restricted CaST labeling. - Viability: DRAQ7 staining at DIV19 comparing neurons expressing CaST, single-fragment control, and fixed-permeabilized positive control. In vivo psilocybin experiments in mice: - Stereotaxic AAV2/1-Synapsin-CaST injections into prelimbic mPFC; 1 week expression. - Labeling session in freely moving mice: i.p. biotin (24 mg/kg) + saline or biotin + psilocybin (3 mg/kg). Recorded head-twitch responses (HTRs) during first 20 min. Euthanized 1 h post-injection; processed 60-µm mPFC sections; SA-647 staining and imaging. - Quantification: SA-647 and GFP per neuron; SA-647/GFP ratios; fraction of SA-647+ neurons (thresholded by saline control 90th percentile). Correlated HTR counts with SA-647+ neuron density and mean SA-647/GFP per FOV. - Controls: Non-CaST mice injected with biotin + psilocybin showed no SA-647 labeling. - Validation: 2-photon Ca2+ imaging with GCaMP6f via GRIN lens in head-fixed mice following saline vs psilocybin to confirm increased neuronal activity. - Comparison with c-Fos: Parallel c-Fos immunostaining in mPFC and somatosensory cortex (SSC) after saline vs psilocybin to assess sensitivity versus CaST. Imaging and analysis: - Epifluorescence and confocal microscopy; automated cell segmentation (cell-segm), Fiji/ImageJ and MATLAB scripts for SA-647/GFP quantification; ROC analyses; Pearson correlations; ANOVAs and nonparametric tests as appropriate. Western blots with SA-HRP and anti-V5. Key parameters: - Biotin 50 µM (cells, neurons); HEK: CaCl2 up to 10 mM with 1 µM ionomycin; Neurons: KCl 30 mM; DOI 10 µM; DA 10 µM; Optogenetics: orange light pulses (20 Hz, 5 ms) cycled (2 s on/4.5 s off) for 30 min; In vivo: biotin 24 mg/kg i.p.; psilocybin 3 mg/kg i.p.; fixation and SA-647 staining 1 h post-injection.

- CaST design and performance: - CaST enables rapid, Ca2+-dependent biotinylation only when exogenous biotin is present. Labeling detectable within 10 min; increased labeling with longer durations up to ~1 h (saturation), and with higher Ca2+ concentrations. - Optimal configuration: membrane-tethered CD4-sTb(C)-M13-GFP + cytosolic CaM-V5-sTb(N); best transfection ratio 5:2. Bicistronic CaST-IRES outperformed P2A (SBR ~5-fold vs 2.7-fold). - Reversibility: pre-stimulation with Ca2+, 10-min wash, then biotin yielded no labeling, indicating temporal resolution on the order of 10 min (ANOVA with Tukey post hoc: significant difference vs simultaneous biotin + Ca2+; P = 1.3 × 10^-10). - Ca2+ titration: SA-647/GFP increased linearly from 2.5–10 mM CaCl2 (Pearson R = 0.99, P = 0.001). - ROC performance in HEK293T: AUC = 0.87 (non-IRES), AUC = 0.93 (IRES) distinguishing +Ca2+ vs -Ca2+. - Comparison to FLICRE: - CaST readout increased immediately after 15-min stimulation and remained above control at all time points (0–8 h delays). FLICRE reporter expression differences emerged only after ~6 h. - Normalized SBRs over 0–8 h: CaST approximately ×3.1, ×2.7, ×2.4, ×2.0, ×1.6; FLICRE approximately ×1.2, ×0.9, ×1.1, ×1.9, ×2.6. At higher expression (48 h post-transfection), FLICRE suffered nonspecific background leading to no clear group differences up to 24 h, whereas CaST maintained immediate discrimination. - Neuronal culture: - 10-min KCl + biotin: ~35% of GFP+ neurons SA-647+ vs ~10% biotin-only; 30-min KCl + biotin: ~65% SA-647+ vs ~10% control. ROC AUC = 0.91 for 30-min condition. - RCaMP2 correlation: higher Ca2+ transient peak heights correlated with greater CaST SA-647/GFP (Pearson R = 0.37, P = 0.035). - Pharmacology: DOI (10 µM) increased CaST labeling; DA (10 µM) did not, consistent with RCaMP2 imaging (DOI and KCl increased activity; DA did not). - Spatial specificity: bReaChES optogenetic stimulation through a 1-mm slit increased SA-647/GFP only within illuminated region; whole-dish KCl produced uniform labeling. - Viability: DRAQ7 staining indicated no overt cytotoxicity in CaST-expressing neurons compared to controls. - In vivo psilocybin labeling and behavior: - Psilocybin (3 mg/kg) increased SA-647 labeling in mPFC neurons vs saline controls; higher mean SA-647/GFP ratios and increased fraction of SA-647+ among GFP+ neurons (~70% vs saline thresholding). - Positive correlation between head-twitch responses (HTRs) and number of SA-647+ neurons/mm^2 during labeling (Pearson R = 0.85, P = 0.03). Mean SA-647/GFP also higher with HTRs. - c-Fos comparison: No increase detected in mPFC by c-Fos after psilocybin (high baseline); SSC showed expected increase, confirming staining quality. CaST detected psilocybin-induced activity increases in mPFC even when normalized to GFP+ counts. - 2P GCaMP6f validation: substantial population of mPFC neurons activated by psilocybin in head-fixed imaging, consistent with CaST results.

CaST directly addresses the need for a rapid, noninvasive, and stable biochemical tag of recent cellular Ca2+ activity. By coupling calmodulin–M13 Ca2+-dependent interaction to split-TurboID reconstitution, CaST functions as a reversible, time-gated integrator whose signal scales with both Ca2+ level and labeling duration. This enables immediate post-labeling readout and compatibility with freely moving animals without fiber optics. In vitro, CaST robustly discriminated activated cells within minutes and outperformed transcriptional integrators that require hours to produce signal. In neurons, CaST captured activity-dependent labeling across brief stimuli, tracked pharmacological specificity (DOI vs DA), correlated with real-time Ca2+ dynamics, and provided spatial precision with targeted optogenetics. In vivo, CaST revealed large-scale activation of mPFC neurons by psilocybin and quantitatively linked cellular activation to HTRs—an outcome difficult with head-fixed imaging or variable IEG markers. These findings highlight CaST’s utility for bridging cellular-resolution activity history with downstream molecular analyses and behaviors inaccessible to optical recording, particularly in deep brain regions.

The study introduces CaST, a Ca2+-activated split-TurboID that rapidly and reversibly biotin-tags proteins in activated cells, enabling immediate, noninvasive readout of recent cellular activity. CaST integrates Ca2+ over user-defined windows controlled by systemic biotin, detects activity within 10 minutes, scales with Ca2+ level and duration, and correlates with behavior in untethered mice. Compared to light-gated and IEG-based reporters, CaST offers deep-tissue scalability and immediate detection, complementing existing tools. Future directions include: brain-wide activity-dependent labeling via systemic gene delivery (e.g., PHP.eB) or transgenic strategies; coupling CaST tagging with spatial omics (MERFISH, STARmap, MALDI-IHC, CITE-seq-like approaches) to profile activated-cell states; using CaST for Ca2+-dependent proximity proteomics via enrichment and mass spectrometry; and refining in vivo protocols (biotin timing, behavior strength, thresholds) for diverse applications.

- Temporal gating relies on systemic biotin windows that are broader than millisecond-second light-gated tools; some behaviors may require sub-minute precision unattainable without light. - Endpoint assay: labeled tissues must be fixed and stained (or lysed), precluding live, longitudinal readout from the same cells. - Does not confer genetic access or enable manipulation of the tagged cells (unlike transcriptional reporters driving opsins). - In vivo use requires careful optimization of biotin timing, stimulus strength, and thresholds with appropriate negative controls. - Although no overt cytotoxicity was observed, prolonged expression and labeling in vivo warrant further cell-physiology and health assessments. - Performance can depend on expression levels; while CaST was robust across conditions tested, very high expression can increase nonspecific background in some systems (notably problematic for protease-based tools).