Light-activated tetanus neurotoxin for conditional proteolysis and inducible synaptic inhibition in vivo

Abstract

The light chain of tetanus neurotoxin (TeNT) is a metalloprotease that potently inhibits synaptic transmission by cleaving the endogenous vesicle fusion protein VAMP2, but its constitutive activity prevents spatiotemporal precision. To address this, we engineered light-activated TeNT (LATeNT) by inserting the light-sensitive LOV domain into an allosteric site and optimizing dynamic range via directed evolution. LATeNT’s activity is undetectable in the dark, but the protease turns on after 10–20 min of weak blue-light exposure to potently inhibit synapses in vivo. Here we show that LATeNT works across multiple brain regions and at long-range axonal projections, with its effects reversible in 24 h. LATeNT enabled us to discover a hippocampal interneuron population that regulates anxiety-like behaviors and demonstrate the importance of postsynaptic endocannabinoid exocytosis for depolarization-induced suppression of inhibition in vivo. Beyond neuroscience, LATeNT regulated endogenous insulin secretion from pancreatic beta cells and converted drug exposure, elevated Ca²⁺ or receptor activation into transgene expression or reporter secretion in HEK293T cells. With a large dynamic range, high light sensitivity and sustained effect, LATeNT enables versatile, spatiotemporally resolved proteolysis across diverse biological systems.

LATeNT is a light-activatable form of tetanus neurotoxin that can be used for spatiotemporally precise inhibition of exocytosis in neurons and other cells. It can inhibit synaptic transmission in vivo or abolish endogenous insulin secretion in pancreatic beta cells in a light-dependent manner, in addition to other applications.

Main

Neurons and many other cell types secrete signaling molecules to mediate cell–cell communication and maintain homeostasis^1,2. A specialized secretion mechanism used by neurons involves SNARE complexes that link synaptic vesicle proteins such as VAMP2 with SNAP25 on the plasma membrane, driving membrane fusion¹. Due to the central and conserved role of VAMP2 and SNAP25 in mammalian biology, bacterial toxins have evolved specific mechanisms to interfere with these proteins, preventing secretion and poisoning the mammalian host. Tetanus neurotoxin (TeNT) from the bacterium Clostridium tetani cleaves VAMP2 via a metalloprotease encoded in its light chain³. This catalytic mechanism makes TeNT extremely potent, with as little as 0.2 ng kg⁻¹ being lethal to humans⁴.

Despite TeNT’s neurotoxicity, its highly specific mechanism⁵, resulting in cleavage of only a single protein in the mammalian proteome, has made TeNT a powerful tool for neuroscience. The 52-kD protease domain of TeNT can be genetically targeted to specific cell populations to silence synaptic transmission and has enabled discovery of functional circuits in spatial learning and social behaviors^6,7. In addition, TeNT and the related toxin BoNT (botulinum neurotoxin from Clostridium botulinum) have been harnessed for delivery of protein cargo across the plasma membrane^8–11, and BoNT has been re-engineered to cleave alternative protein substrates such as PTEN¹².

To fully realize the potential of these toxins as tools for biology and neuroscience, precise spatiotemporal control over their protease activities is needed. A single-chain, light-regulated TeNT that is activated rapidly in specific brain regions during user-selected time windows would enable functional circuit dissection with high precision and minimal toxicity. Proteolysis-mediated silencing by such a construct would be long-lasting yet reversible through the replenishment of endogenous VAMP2 molecules. A simple, single-chain design would also represent an advance over related tools such as photoactivatable BoNT (PA-BoNT)¹³ that exhibited small dynamic range in our hands (Supplementary Text 1).

In this work, we used structure-guided protein engineering and directed evolution to produce a compact (69 kD), single-chain light-activated TeNT (LATeNT). LATeNT enables rapid and reversible control of synaptic transmission in the mouse brain, leading to the discovery of a hippocampal interneuron population that controls anxiety-like behaviors. We also show that LATeNT can modulate endocannabinoid-based retrograde signaling at inhibitory synapses and insulin secretion from pancreatic beta cells. Finally, leveraging LATeNT’s high sequence specificity, synthetic circuits that convert Ca²⁺, drug or receptor activity into gene transcription and reporter secretion were constructed in HEK293T cells that lack endogenous VAMP2 expression. With its simple design and robust performance, we expect LATeNT to have broad utility in neuroscience and synthetic biology.

Results

Design and directed evolution of LATeNT

To engineer LATeNT, we considered multiple strategies for incorporating light-responsive domains into the 52 kD TeNT protease domain (hereafter referred to as TeNT). We opted against splitting TeNT and fusing the fragments to light-controlled heterodimers such as CRY2/CIBN, because reconstituted split enzymes usually have activity far below that of the parental full-length enzymes^14,15, and two-component tools exhibit higher variability than one-component tools¹⁶. Instead, we pursued a single-chain solution (Fig. 1a), similar to the light-regulated proximity labeling enzyme LOV-Turbo that we recently reported¹⁷. We envisioned inserting the light-sensitive LOV domain into a surface-exposed loop of TeNT to ‘clamp’ the loop in the dark state. If the insertion site is allosterically coupled to TeNT’s active site, clamping could inactivate the protease by distorting its active site. Blue light would trigger the release of LOV domain’s C-terminal Jα helix, ‘unclamping’ the loop and restoring activity. If successful, a LOV-containing TeNT would be ~69 kD, monomeric, reversible, easily packaged in an adeno-associated virus (AAV) vector and effective across a range of expression levels.

Fig. 1. Design and directed evolution of LATeNT.

a, Design of LATeNT. hLOV (blue), an engineered light-sensitive LOV domain¹⁷, is inserted between amino acids 357 and 358 of TeNT protease and keeps the protease inactive in the dark. Protease activity is restored upon blue-light (470 nm) stimulation. Active LATeNT inhibits neurotransmitter release by cleaving the endogenous synaptic vesicle fusion protein VAMP2. b, AlphaFold3-predicted structure of LATeNT. TeNT is shown in green, LOV domain in blue, active site residues in yellow, zinc ion in orange and two mutations introduced by directed evolution in red. c, hLOV insertion sites tested in TeNT. The color indicates the result of activity test in d. d, Relative activities of LOV insertion constructs from c. HEK293T cells stably expressing mCherry–myc–VAMP2 reporter (‘HEK293T-VAMP2 reporter cells’) were stimulated with ambient room light for 30 min, then lysed. VAMP2 cleavage was quantified by western blot using an antibody that detects the cleaved form of VAMP2²⁶. LATeNT0 was best (see Supplementary Fig. 1a for western blot). e, Optimization of linkers in LATeNT0 produces LATeNT1. n:m indicates linkers with n and m residues N-terminal and C-terminal to the inserted LOV domain, respectively. m = −1 to −4 means that LOV was truncated on its C-terminal side by 1–4 residues (see Supplementary Fig. 1b for western blot). f, Directed evolution of LATeNT in yeast. Cells were stimulated with light (1 h or overnight) or kept in the dark, then sorted by YFP/mCherry ratio. The table shows conditions for each round of selection. TF is the LexA-VP16 transcription factor, which drives YFP reporter expression. g, FACS analysis of template (LATeNT1), initial library and post-round 5 library. Percentages indicate fraction of LATeNT-expressing cells in quadrant Q2. h, Comparison of LATeNT1 and final evolved LATeNT (‘C16’) in HEK293T-VAMP2 reporter cells after 30 min of 470 nm light stimulation. Cleaved VAMP2 is the N-terminal cleavage product of mCherry–myc–VAMP2 reporter (41.5 kD, cleaved into 36.8 and 4.7 kD fragments), detected using VAMP/B/1148 antibody²⁶. Untrans., untransfected. Data are mean ± s.d. (n = 3). See Supplementary Fig. 6a for western blots of biological replicates.

Source data

To engineer LATeNT in this manner, we selected 14 positions in 12 surface-exposed loops of TeNT based on its 2.3 Å crystal structure¹⁸ (Fig. 1c) and inserted hLOV, an engineered LOV domain with tighter dark-state caging¹⁹. These constructs were expressed in HEK293T cells stably expressing mCherry–myc–VAMP2 (‘HEK293T-VAMP2 reporter cells’). Of the 14 insertion sites tested, one site (357/358) showed ~1.5-fold increase in VAMP2 cleavage after 30 min of room light compared with dark controls (Fig. 1c,d and Supplementary Fig. 1a). We designated this construct LATeNT0. To increase the light/dark signal ratio, we varied the lengths of linkers flanking the inserted hLOV domain. One amino acid truncation of the C-terminal Jα helix substantially improved dynamic range, perhaps by improving allosteric coupling between hLOV and TeNT’s active site (Fig. 1e and Supplementary Fig. 1b). Our optimized ‘LATeNT1’ exhibited an improved light/dark activity ratio of 6.7 (Fig. 1h).

To further reduce the dark-state activity of LATeNT1 (Fig. 1h), which we were concerned would produce background on long-term expression in vivo, we performed directed evolution. Yeast cells (Saccharomyces cerevisiae) are genetically tractable and well-suited for screening large libraries by fluorescence-activated cell sorting (FACS). To convert LATeNT activity into a FACS-sortable signal, we tethered VAMP2 to the plasma membrane and fused its N-terminus to the transcription factor LexA–VP16. VAMP2 cleavage by TeNT frees LexA-VP16 for translocation to the nucleus, where it can drive YFP reporter expression (Fig. 1f and Supplementary Fig. 2c).

We created a library of LATeNT1 variants by error-prone PCR and fused the library to mCherry to enable quantification of expression levels. Positive selection was performed by exposing yeast cultures to room light during overnight induction, waiting 8 h for YFP reporter expression, then enriching cells with high YFP/mCherry ratio by FACS (Fig. 1f). After amplification of sorted cells, we performed negative selection by culturing in the dark and retaining cells with low YFP/mCherry ratio (Fig. 1f). In total, we performed five rounds of selection—three positive rounds and two negative rounds (Fig. 1f and Supplementary Fig. 1c). Analysis of reamplified cells by flow cytometry showed improvements in both (+)light activity and (−)light background over the course of evolution (Fig. 1g and Supplementary Fig. 1d).

After the fifth round of selection, we analyzed 20 enriched clones (Supplementary Fig. 2a,b). Clone 16 (C16), with one mutation in the TeNT protease (K34R) and one mutation in the LOV domain (D397N) (Fig. 1b), showed substantially lower dark-state activity than LATeNT1 while maintaining comparable (+)light activity in yeast (Supplementary Fig. 2c). This trend was recapitulated in HEK293T-VAMP2 reporter cells, where the light/dark activity ratio was 97.3, compared with 6.7 for LATeNT1—primarily due to a reduction in dark-state activity (Fig. 1h and Supplementary Fig. 2a,b). C16 was named LATeNT and used for subsequent experiments.

Characterization of LATeNT in HEK293T cells and cultured neurons

To compare LATeNT’s catalytic efficiency with wild-type TeNT, we incubated TeNT- or LATeNT-containing HEK293T lysates with VAMP2-containing lysates and monitored VAMP2 cleavage over time. Western blot shows that LATeNT, in the presence of light, retains ~36% of wild-type TeNT’s activity (Fig. 2a). To validate LATeNT’s mechanism, we introduced point mutations in its protease and LOV domains. E233A in the protease domain²⁰ and C415A in LOV²¹ abolished activity, whereas the I504E mutation in the LOV domain rendered LATeNT constitutively active (Extended Data Fig. 1a).

Fig. 2. Characterization of LATeNT in HEK293T cells and cultured neurons.

a, Comparison of TeNT and LATeNT cleavage rates. HEK293T lysates expressing TeNT or LATeNT were added to lysates expressing recombinant VAMP2 for 10–40 min. LATeNT samples were stimulated with 470 nm light during in vitro reaction. Data are mean ± s.d. (n = 3). See Supplementary Fig. 6b for western blots of biological replicates. b, Kinetics of LATeNT turn-off after light removal. HEK293T-VAMP2 cells expressing LATeNT or LATeNT* (V381L mutant of LATeNT) were stimulated with 470 nm light for 30 s to 2 h, then kept in the dark for the remaining time, for a total of 2 h. Data are mean ± s.d. (n = 3) (see Extended Data Fig. 1d and Supplementary Fig. 6c for western blots). c, mCherry–LATeNT cleaves endogenous VAMP2 in cultured neurons. After stimulation with 470 nm light for 30 min, rat neurons were fixed and stained with VAMP/B/151 antibody and anti-Bassoon antibody (presynaptic marker). Yellow arrows indicate colocalization between Bassoon and cleaved VAMP2 puncta along LATeNT-expressing (white V5+) processes. Red arrows point to Bassoon puncta in LATeNT-negative cells lacking cleaved VAMP2 staining. See Extended Data Fig. 1e for additional fields of view. d, Endogenous VAMP2 is replenished in neurons 24 h after LATeNT-catalyzed cleavage. After LATeNT stimulation with 470 nm light for 1 h, neurons were allowed to recover for 0–24 h, then lysates were analyzed by western blotting. Uninfected (Uninf.) or mCherry-TeNT-expressing (TeNT) neurons were used as controls. Three biological replicates were performed (see Supplementary Fig. 6d for additional replicates).

Source data

Extended Data Fig. 1. Characterization of LATeNT in HEK293T cells and cultured neurons.

(A) Western blot characterizing effect of point mutations in LATeNT on VAMP2 cleavage in HEK293T-VAMP2 cells. Point mutations that inactivate protease activity (E233A²⁰), prevent opening of the LOV domain (C415A²¹), or prevent closing of the LOV domain (I504E²²) were introduced. Light stimulation time was 30 min. This experiment was performed 2 times with similar results. (B, C) LATeNT activation with (B) different light intensities or (C) different duty cycles. In (B), HEK293T-VAMP2 cells were transfected with LATeNT and stimulated with varying intensities of ‘continuous’ 470 nm blue light for 30 min. In (C), cells were stimulated with 1.0 mW/cm² 470 nm blue light with different duty cycles (2 s on with 0 to 38 s off) for 30 min. (D) Western blots showing activities of LATeNT and LATeNT* with increasing duration of light stimulation. HEK293T-VAMP2 cells stably expressing LATeNT and LATeNT* were stimulated with 470 nm blue light for 30 s to 2 h, and were kept in the dark for the remaining time. Western blot of biological replicates in Supplementary Fig. 6C. (E) Additional fields of view showing cleavage of endogenous VAMP2 in cultured neurons.

Source data

To test LATeNT’s reversibility, we stimulated cells with blue light for 30 s to 2 h, then kept them in the dark for the remainder of the 2 h experiment (Fig. 2b, left). VAMP2 cleavage increased linearly with light stimulation time (Fig. 2b and Extended Data Fig. 1d), suggesting that LATeNT activity shuts off within minutes of light removal, similar to other LOV-based tools^17,22. We also made a point mutant of LATeNT, named LATeNT*, with a V381L mutation in LOV that slows LOV domain closing by >75-fold (τ = 72 min instead of τ = 55 s (ref. ²³)). In the same reversibility experiment, VAMP2 cleavage by LATeNT* reached saturation regardless of light stimulation time (Fig. 2b and Extended Data Fig. 1d), suggesting that LATeNT* remains active (for over 60 min) even after light is removed.

In neurons, endogenous VAMP2 is localized primarily in presynaptic vesicles in axon terminals^24,25. To test if LATeNT can traffic to these sites and cleave endogenous VAMP2 in situ, we expressed LATeNT in cultured neurons via AAV infection. Confocal imaging shows the presence of LATeNT along processes and at puncta containing the presynaptic marker Bassoon (Fig. 2c and Extended Data Fig. 1e). A total of 30 min of light stimulation produced puncta of cleaved VAMP2 (visualized by an antibody that selectively stains cleaved VAMP2²⁶), which colocalized with Bassoon on LATeNT-expressing processes but not on processes lacking LATeNT expression (Fig. 2c). Western blotting showed that 1 h light depleted endogenous VAMP2 in cultured neurons by 78% ± 3% (mean ± s.d.; Fig. 2d and Supplementary Fig. 6d), which was fully restored over ~16–24 h in the dark (Fig. 2d).

LATeNT inhibits synaptic transmission in slice and in vivo

To assess the functional effects of LATeNT, we first transduced cultured hippocampal neurons and measured miniature excitatory postsynaptic currents (mEPSCs) (Fig. 3a). Light stimulation resulted in significant reduction of both mEPSC frequency and amplitude, comparable to that caused by wild-type TeNT expression (Fig. 3b–d). We then tested LATeNT in the mouse brain, injecting mCherry–LATeNT AAVs into hippocampal CA3 and measuring evoked excitatory postsynaptic currents (eEPSCs) in CA1 neurons by simulating Schaffer-collateral axons (Fig. 3e and Extended Data Fig. 2a). Acute brain slices exposed to light showed a significant reduction in both AMPAR- and NMDAR-mediated eEPSC amplitudes compared with dark controls (Fig. 3f–h and Extended Data Fig. 2b,c). The reduction in excitatory synaptic strength in the Schaffer-collateral pathway was accompanied by reduced neurotransmitter release probability (P_r), indicated by increased paired-pulse ratios (PPR) after light exposure (Fig. 3i,j).

Fig. 3. LATeNT inhibits synaptic transmission.

a, mEPSC recording scheme in cultured hippocampal neurons. b, Representative mEPSC traces recorded from uninfected and LATeNT-expressing neurons kept in the dark (light gray) or exposed to 0.55 mW cm⁻² LED light for 30 min (blue). c,d, A quantification of mEPSC frequencies (c) and amplitudes (d) in b. Data are mean ± s.e.m. (uninfected, n = 19; LATeNT(dark), n = 21 (P = 0.7042/0.061 for frequency/amplitude); LATeNT(light), n = 26 (P = 0.0002/0.0168 for frequency/amplitude); TeNT, n = 15 (P ≤ 0.0001/0.0004 for frequency/amplitude); n.s., not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; two-sided ANOVA with Tukey’s post hoc test). e, Scheme for measuring synaptic connectivity in acute hippocampal slices. CA3 presynaptic fibers were stimulated while monitoring CA1 postsynaptic responses. Ambient room light (~0.15–0.2 mW cm⁻²) was supplied for 1–2 h. f–h, Representative AMPAR-EPSC traces (f), summary of EPSC amplitudes versus CA3 stimulation current (g) and summary of fitted linear input–output (I–O) slopes (h). Excitatory synaptic strength, measured using I–O curves, from CA3–CA1 synapses are measured. Data are mean ± s.e.m (n = neurons/mice; dark, n = 14/4; light, n = 14/4; **P < 0.01, ****P < 0.0001; two-sided Mann–Whitney U test). For g, P = 0.0082 (20-μA stimulus intensity) or <0.0001 (40, 60, 80, 100 μA stimulus intensity). For h, P = <0.0001. i, Representative paired-pulse EPSC traces recorded at CA3–CA1 synapses with a 50-ms interstimulus interval (ISI). j, Quantification of PPRs at CA3–CA1 synapses across different ISIs (50, 100, 200 and 500 ms). Data are mean ± s.e.m. (n = neurons/mice; dark, n = 13/4; light, n = 12/4; **P < 0.01, ****P < 0.0001; two-sided Mann–Whitney U test). P = 0.0012/<0.0001/0.0037/0.1095 for 50/100/200/500 ms ISI. k, EPSC recording during continuous blue-light illumination (473 nm, 0.5 mW cm⁻²) of brain slices expressing mCherry–LATeNT (LATeNT) or mCherry only. Data are mean ± s.e.m. (n = neurons/mice; mCherry(light), n = 4/3; LATeNT(dark), n = 4/3, LATeNT(light), n = 4/3; *P < 0.05, LATeNT(dark) versus LATeNT(light) from the indicated time points; two-way repeated measures ANOVA with Tukey’s post hoc test). l, Recovery kinetics of synaptic inhibition by LATeNT and PdCO. Left: experimental schematic. Cultured hippocampal neurons expressing LATeNT or PdCO were stimulated with blue (470 nm, 30 min) or UV (395 nm, 5 min) light, respectively, with eEPSCs subsequently measured at indicated times. Right: summary plots of eEPSC amplitudes over time. Data are mean ± s.e.m. LATeNT: n = 20/17/16/16/19 neurons; PdCO: n = 20/17/17/16/19 neurons for dark/5 min/1 h/8 h/24 h, respectively; from two to three independent culture batches. **P < 0.01, ***P < 0.001, ****P < 0.0001; Kruskal–Wallis test. Pairwise comparisons at each time point performed using multiple Mann–Whitney tests with Holm–Šidák correction (^#P < 0.05, ^##P < 0.01, ^####P < 0.0001). m, LATeNT mediates long-range synaptic inhibition. Scheme showing AAV injection (LATeNT + ChrimsonR) into vHPP. A total of 4 weeks post injection, oEPSCs in mPFC slices were elicited by red-light stimulation (590 nm, 10 mW mm⁻², 30 ms) of vHPP projections. LATeNT was activated by ambient room light before red-light stimulation. n, Representative monosynaptic oEPSC traces recorded under TTX + 4-AP conditions in response to red light (red lines) in neurons expressing ChrimsonR ± LATeNT. o, A quantification of oEPSC amplitudes in n. Data are mean ± s.e.m. (ChrimsonR, n = 15 cells from 5 mice; ChrimsonR + LATeNT, n = 15 cells from 4 mice (P = 0.0113); *P < 0.05, two-sided Mann–Whitney U test). p, Scheme testing LATeNT spatial specificity. AAVs expressing mCherry–LATeNT were injected unilaterally into CA3, with optical fibers implanted into contralateral CA1. Mice were subjected to 473 nm light (30 min; 2 s on/10 s off) 3 weeks post injection. q, mCherry–LATeNT expression in CA3. The white boxes indicate quantification areas for r. r, Representative images of ipsilateral and contralateral CA1 stained for the C-terminal fragment of cleaved VAMP2 (cVAMP2, green). Yellow arrows show red-green colocalization. s, A quantification of mCherry-colocalized cVAMP2 puncta in CA1. Data are mean ± s.e.m. (n = 4 mice per group; 3 to 4 brain sections quantified per mouse; ****P < 0.0001, two-sided Student’s t-test).

Source data

Extended Data Fig. 2. LATeNT inhibits synaptic transmission.

(A) Representative brain section showing mCherry-LATeNT expression in the hippocampal CA3 after performing the experiment shown in Fig. 3e. Image shows a section obtained from the Bregma −2.0 mm region. (B, C) Representative traces (B) and summary graphs (C) of NMDAR/AMPAR-EPSC ratios at CA3-CA1 synapses. AMPAR-EPSCs were recorded at -70mV in the presence of picrotoxin, and NMDAR-EPSCs were then recorded at 40 mV n = neurons/mice; dark, n = 10/4; light, n = 11/4; p = 0.4262, Mann-Whitney U test. (D) Paired plots of EPSC amplitudes averaged over the first 5 min (before) and last 5 min (after). Each pair of line-connected dots represents an individual neuron (*p < 0.05; paired t-test). p = 0.8812 (mCherry+light), 0.4797 (LATeNT+Dark), 0.0444 (LATeNT+Light). (E, F) Representative eEPSC traces of (E) LATeNT-expressing and (F) PdCO-expressing neurons at each time point. Quantification of replicates are provided in Fig. 3l. (G, H) Effects of increasing light intensity on EPSC amplitudes. mCherry-LATeNT-expressing slices were illuminated with blue light at indicated intensities for 30 min, while electrically stimulating CA3 fibers at 100 µA. (G) Representative EPSC traces at each light intensity. (H) Summary plot of EPSC amplitudes as a function of light intensity. Data are presented as mean ± SEM (At each light intensity (0, 0.01, 0.03, 0.05, 0.1 mW/cm²), n = neurons/mice was as follows: Dark, 10/2 at all intensities; Light, 10/2, 9/2, 10/2, 9/2, and 9/2, respectively; *p < 0.05, ***p < 0.001, ****p < 0.0001; Mann-Whitney U test). p = 0.9118/0.8421/0.0147/ < 0.0001/0.0004 for 0/0.01/0.03/0.05/0.10 mW/cm² light intensity. (I) Representative image showing robust mCherry-LATeNT and ChrimsonR expression in the vHPP injection site, following experimental scheme in Fig. 3m. (J) Schematic illustration of the experimental workflow for optogenetically evoked EPSC (oEPSC) recordings from long-range vHPP→mPFC projections. Recordings were performed sequentially under (1) control (no drug) conditions, followed by (2) bath application of tetrodotoxin (TTX), and subsequently (3) TTX plus 4-aminopyridine (4-AP) to isolate monosynaptic, terminal-driven synaptic responses. (K) Representative oEPSC traces recorded from mPFC neurons in response to red light stimulation of ChrimsonR-expressing vHPP axon terminals under the three conditions shown in (J): (1) Control (no drug), (2) TTX, and (3) TTX + 4-AP. Traces are shown for neurons expressing ChrimsonR alone and ChrimsonR together with LATeNT. Red bars indicate red light stimulation. (L) Mice were injected with AAVs expressing mCherry-LATeNT into the vHPP at 6 weeks of age, followed by optical fiber implantation in the mPFC at 7 weeks and light stimulation at 9 weeks prior to brain slice preparation and immunohistochemistry. Representative images (left) and summary graphs (right) of mPFC regions stained with VAMP/B/151 antibody that detects C-terminal fragment of cleaved VAMP2 (green). Yellow arrows indicate colocalization of mCherry with cleaved VAMP2 puncta. Data are presented as mean ± SEM (n = 8-9 from 3 mice per group; p = 0.0014; **p < 0.01, Mann-Whitney U test).

Source data

To determine the kinetics of LATeNT-mediated inhibition, we measured eEPSCs in CA1 neurons of acute brain slices prepared as above, under continuous blue-light illumination. We observed a gradual decrease in eEPSC amplitudes only in LATeNT-expressing slices, noticeable within 5 min and reaching a maximum after 20 min (Fig. 3k). These averaged measurements were validated by paired comparison of EPSC amplitudes from individual cells (Extended Data Fig. 2d).

We also measured the duration and recovery time of LATeNT-mediated inhibition in hippocampal neuron cultures expressing mCherry–LATeNT. We found that evoked EPSCs were reduced for 8 h following 30 min of blue-light exposure and fully recovered after 24 h (Fig. 3l and Extended Data Fig. 2e). For comparison, we used the same experimental scheme to evaluate PdCO²⁷, a recently reported bistable rhodopsin activated by 405 nm ultraviolet (UV) light. PdCO-mediated eEPSC inhibition (after 5 min of UV activation) lasted for 1 h and recovered after 8 h (Fig. 3l and Extended Data Fig. 2f). Quantitation showed that inhibition at the 1 h time point was significantly stronger for LATeNT than for PdCO (Fig. 3l). Thus, LATeNT could be beneficial for experiments requiring either more sustained or more complete synaptic inhibition.

To characterize LATeNT’s light intensity requirements, we exposed brain slices to blue light ranging from 0 to 0.1 mW cm⁻² in power. We found that eEPSC amplitudes were significantly reduced with power as low as 0.03 mW cm⁻² and the effect saturated at 0.05 mW cm⁻² (Extended Data Fig. 2g,h). Similarly, LATeNT was activated by very low intensity light in HEK293T cells (Extended Data Fig. 1b,c). Although LATeNT requires a higher cumulative light dose than other optogenetic tools that work on the subsecond timescale, the low-power requirement is helpful for minimizing phototoxicity to tissues (Supplementary Text 2).

We next tested LATeNT’s ability to inhibit synaptic transmission in long-range projections. LATeNT was co-expressed with the red-shifted opsin ChrimsonR in the ventral hippocampal projection population (vHPP), which projects to the medial prefrontal cortex (mPFC) (Fig. 3m). Red light stimulation of vHPP neurons evoked EPSCs in mPFC neurons. Under tetrodotoxin (TTX) and 4-aminopyridine (4-AP) treatment to isolate monosynaptic transmission, ambient room light activation of LATeNT robustly inhibited these evoked EPSCs (Fig. 3n,o and Extended Data Fig. 2i–k). Immunofluorescence staining showed that LATeNT was present at vHPP-mPFC synapses, indicating successful long-range trafficking, and light-dependent increases in cleaved VAMP2 puncta were observed at these sites (Extended Data Fig. 2l).

To test the spatial specificity of light-based activation, we performed an experiment in which mCherry–LATeNT was unilaterally expressed in CA3 neurons, which project to both ipsilateral and contralateral CA1²⁸ (Fig. 3p,q). When light was delivered to the contralateral CA1 only via optical fiber, this led to an increase in cleaved VAMP2 levels in the contralateral CA1 but not in the non-illuminated ipsilateral CA1 (Fig. 3r,s). We also asked whether somatic illumination in acute brain slices can produce VAMP2 cleavage at synaptic termini. Our results show that VAMP2 cleavage at terminals occurs mainly through local protease activation (Extended Data Fig. 3), although cleaved VAMP2 generated at the soma can be transported to synaptic sites over many hours (Supplementary Text 3). Although long-term (24 h) eEPSC recovery was unimpaired in cultured neurons (Fig. 3l), we did not fully characterize the effects of soma-derived cleaved VAMP2 on the properties of distal synapses over long timescales.

Extended Data Fig. 3. Somatic LATeNT activation can lead to time-dependent accumulation of cleaved VAMP2 at axon terminals.

(A) Experimental scheme. AAVs expressing mCherry-tagged LATeNT were injected unilaterally into the hippocampal CA3 region, and an optical fiber was implanted above the same (ipsilateral) CA3. Three weeks after injection, mice were subjected to 473 nm blue light illumination (30 min; 2 s on/10 s off). Brain perfusion was performed either immediately after illumination (0 hr post-light) or following a 6-hour post-illumination interval (6 hr post-light), prior to immunohistochemical analysis. (B, D) Representative images of ipsilateral CA3 under no-light and light-illumination conditions, and downstream CA1 from the ipsilateral and contralateral hemispheres following ipsilateral CA3 illumination. Panels show data collected at 0 hr (B) and 6 hr (D) post-light. Cleaved VAMP2 (cVAMP2) is shown in green, and mCherry-LATeNT in red. (C, E) Quantification of mCherry-colocalized cleaved VAMP2 puncta corresponding to the 0 hr (C) and 6 hr (E) post-light conditions. Data are presented as mean ± SEM (n = 3–4 mice per group; for each mouse, 3 brain sections were quantified and averaged; *p < 0.05, **p < 0.01, Student’s t-test). For (C), p = 0.0406 (Ipsi-CA3), 0.3307 (Ipsi-CA1). For (E), p = 0.0296 (Ipsi-CA3), 0.0014 (Ipsi-CA1), 0.0362 (Contra-CA1).

Source data

Collectively, our experiments show that LATeNT can potently inhibit synaptic transmission in slice and in vivo, with detectable onset in ~5 min, saturation in 20 min, continued inhibition for ~8 h and reversal in ~24 h. Inhibition can be spatially controlled with low-power blue light and is effective in long-range projections.

LATeNT-mediated synaptic inhibition can alter mouse behaviors

Mice that explore enriched environments (EE) show greater neuronal plasticity and activity than mice housed in standard environments (SE), evidenced by higher c-Fos expression²⁹. To test if LATeNT can manipulate this endogenous response to behavior, we injected AAVs bilaterally into hippocampal CA1 of adult mice and housed them in either SE or EE (Fig. 4a,b and Extended Data Fig. 4a). EE mice received either no light or blue light for 30 min in the dorsal subiculum (dSub) via an implanted fiber, before each of five 8 h-long EE sessions (Fig. 4a). As expected, EE exposure robustly increased c-Fos expression in dSub compared with mice housed in SE (Fig. 4c,d). However, this effect was abolished in mice that received light before EE (Fig. 4c,d), suggesting that LATeNT-mediated inhibition of synaptic transmission from CA1 to dSub suppressed EE-driven c-Fos expression. Light-dependent activation of LATeNT was also confirmed by staining for cleaved VAMP2 (Extended Data Fig. 4b,c). In the auditory cortex (AC), which does not receive direct inputs from CA1 (Extended Data Fig. 4a), light had no effect on EE-driven c-Fos expression (Extended Data Fig. 4d,e). These results demonstrate that LATeNT can suppress experience-driven neuronal activity by inhibiting synaptic transmission from defined inputs, thereby providing a causal link between specific circuit connections and the gene expression programs that underlie behavioral experience.

Fig. 4. LATeNT identifies causal role for CA1 SST⁺ interneurons in anxiety-like behaviors.

a, Experimental scheme for testing LATeNT effect on c-Fos expression after exposure to EE. A total of 3 weeks after AAV injection, 473 nm light was delivered to the dSub for 30 min (1.0 mW cm⁻², 2 s on, 10 s off), after which mice were provided EE conditions for 8 h per day for 5 consecutive days. Sections were stained with anti-c-Fos antibody. b, Hippocampal CA1 neurons expressing mCherry–LATeNT (red) projecting into dSub. DG, dentate gyrus. c, Representative images of c-Fos immunostaining in the dSub region. d, A quantitative analysis of the number of c-Fos–expressing cells per unit area in mice under SE (dark gray), EE (light gray) or EE + light (blue) conditions. Data are presented as mean ± s.e.m. (n = 3 mice each after averaging data from 4 sections/mouse; **P < 0.01, ***P < 0.001; two-sided ANOVA with Tukey’s post hoc test). P = 0.017 (SE versus EE), 0.0005 (EE versus EE + light), 0.3467 (SE versus EE + light). e, Experimental schematic for testing the role of SST⁺ interneurons in anxiety-like behaviors. A total of 3 weeks after AAV injection, the first elevated plus maze (EPM) test was performed. The next day, the second EPM test was performed 6 h after 473 nm light was delivered to the CA1 through an optical fiber for 30 min (2 s on/10 s off). The Third EPM test was performed 1 week later. f, FlpO-mediated recombination of AAV-delivered genes. Frt, Flp recognition target sequence. g, Representative image showing mCherry–LATeNT expression in hippocampal CA1 SST interneurons. SO, stratum oriens, SP, stratum pyramidale. h, Representative top-view heat maps of time spent in open versus closed arms of the EPM test. Control mice do not express LATeNT but receive light stimulation. i, Bar graphs presenting the time spent in open arms across trials. Data are presented as mean ± s.e.m. (n = 8 mice; P = 0.0042; **P < 0.01; two-sided Mann–Whitney U test). j, Before and after graphs presenting the time spent in open arms across trials. Each pair of line-connected dots represents an individual mouse. A total of 6 to 8 mice were analyzed per condition (*P < 0.05; Friedman test). P = 0.0733 (control First EPM versus Second EPM), >0.999 (control Second EPM versus Third EPM), 0.0179 (LATeNT First EPM versus Second EPM), 0.0373 (LATeNT Second EPM versus Third EPM).

Source data

Extended Data Fig. 4. LATeNT inhibits EE-driven c-Fos expression.

(A) Representative image showing mCherry-LATeNT expression in the hippocampal CA1. Scale bar = 500 μm. AC, auditory cortex; DG, dentate gyrus; dSub. dorsal subiculum. (B) Representative images of dSub regions from mice injected with AAVs expressing mCherry-LATeNT and analyzed via immunofluorescence staining with an anti-cleaved VAMP2 (green) antibody. Yellow arrows mark colocalization of mCherry with cleaved VAMP2 puncta. Scale bar = 20 μm. (C) Summary graphs quantifying cVAMP2⁺mCherry⁺ puncta density in mice under SE (dark gray), EE (light gray), or EE+Light (blue) conditions. Data are presented as mean ± SEM (n = 8–9 from 3 mice per group; *p < 0.05, ****p < 0.0001; ANOVA with a non-parametric Kruskal–Wallis test). p = 0.5046 (SE vs EE), <0.0001 (SE vs EE+Light), 0.0119 (EE vs EE+Light). (D) Representative images of c-Fos immunostaining in the auditory cortex (AC; control region). Scale bar = 20 μm. (E) Quantitative analysis of data from experiment in (D). Data are presented as mean ± SEM (n = 3 mice each after averaging data from 4 sections/mouse; *p < 0.05; ANOVA with Tukey’s post hoc test). p = 0.0223 (SE vs EE), 0.0293 (SE vs EE+Light), 0.9697 (EE vs EE+Light).

Source data

Next, we used LATeNT to examine the causal relationship between a specific hippocampal microcircuit and anxiety-like behaviors. Previously, our lab found that conditional knockout of NPAS4, an activity-dependent transcription factor controlling GABAergic synapse development in hippocampal CA1 SST⁺ (somatostatin-positive) interneurons increases anxiety-like behaviors³⁰. However, because NPAS4 regulates multiple pathways including development³¹, it was unclear which downstream process drives this behavioral change. We used LATeNT to explore the hypothesis that SST⁺ GABAergic interneuron activity is causally linked to anxiety-like behaviors in mice.

We expressed mCherry–LATeNT or mCherry alone in SST⁺ GABAergic interneurons using Flp-dependent AAVs (Fig. 4e–g). A total of 3 weeks after AAV injection, we monitored anxiety-like behaviors by performing elevated plus maze (EPM) tests. In the first session, performed without light, both control and LATeNT groups showed comparable anxiety-like behaviors, spending a similar amount of time in open versus closed arms (First EPM; Fig. 4h–j). In the second session, following blue-light stimulation via CA1-implanted fibers, only the LATeNT group showed a significant decrease in time spent in the open arms, consistent with increased anxiety (Second EPM; Fig. 4h–j). To determine if this effect could be reversed, we performed a third session 1 week later and found that the behavior of LATeNT mice was restored to that of control mice (Third EPM; Fig. 4h–j).

To support these findings, we performed additional anxiety-related behavioral tests, including light/dark box test, sucrose preference test and open-field test (Extended Data Fig. 5a). In the light/dark box test, mice naturally prefer the dark box and enter the light compartment more as their anxiety levels decrease³². We observed that LATeNT activation led to a significant reduction in both time spent in and frequency of entries into the light compartment (Extended Data Fig. 5b–d), consistent with increased anxiety. In the sucrose preference test, performed under non-deprived, habituated conditions, control mice exhibited a robust preference for sucrose over water, whereas LATeNT-activated mice showed no sucrose preference (Extended Data Fig. 5e), also consistent with a higher level of anxiety. The open-field test was used as a control to show preservation of general locomotor and exploratory activity; LATeNT-mediated inhibition of SST⁺ interneurons did not alter total distance moved, number of entries into the center zone or time spent in the center zone (Extended Data Fig. 5f–i). Finally, we used electrophysiological recordings on acute brain slices to confirm that LATeNT functionally suppresses synaptic transmission from SST⁺ interneurons to CA1 neurons in a light-dependent manner (Extended Data Fig. 6). Taken together, our results show that suppression of GABAergic synaptic transmission in CA1 SST⁺ interneurons increases anxiety-like behaviors and reduces hedonic preference without affecting general locomotor activity.

Extended Data Fig. 5. Modulation of anxiety-like and hedonic behaviors by LATeNT-mediated inhibition of hippocampal CA1 SST⁺ interneurons.

(A) Experimental scheme. AAVs expressing fDIO-mCherry-LATeNT or fDIO-mCherry were injected into the hippocampal CA1 of Sst-IRES-FlpO mice, and optical fibers were implanted into the hippocampal CA1. Three weeks after injection, open-field test (OFT), light–dark transition test (LDT), and sucrose preference test (SPT) were performed at 3-day intervals, with 30 min of 473 nm light stimulation (2 sec on/10 sec off) delivered immediately prior to each behavioral test. (B) Representative track images in the light chamber during the LDT. (C, D) Time spent in the light chamber (C, p = 0.0421) and number of entries into the light chamber (D, p = 0.0465). Data are presented as mean ± SEM (control, n = 12; LATeNT, n = 15; *p < 0.05; Student’s t-test). (E) Preference (%) for water (W) and sucrose (S) during the SPT in the control (left) and LATeNT (right) groups. Data are presented as mean ± SEM (control, n = 12 (p = 0.0056); LATeNT, n = 11 (p = 0.352); **p < 0.01; Student’s t-test; n.s., not significant). (F) Representative track images in the OFT. (G-I) Total distance moved (G, p = 0.294), number of entries into the center zone (H, p = 0.216), and time spent in the center zone (I, p = 0.3309). Data are presented as mean ± SEMs (Control, n = 12; LATeNT, n = 15; Student’s t-test; n.s., not significant.).

Source data

Extended Data Fig. 6. Electrophysiological validation of LATeNT-mediated inhibition of SST+ interneurons.

Hippocampal CA1 SST⁺ interneurons innervate the distal dendritic compartment of pyramidal neurons in the CA1 (A) Schematic showing recording at CA1 pyramidal neurons with stimulating electrode at SLM layer (dendritic stimulation). SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatum; SLM, stratum lacunosum-moleculare. (B, C) Representative dendritic eIPSC traces, average eIPSC I-O curve (B), and average eIPSC I-O slope (C) from hippocampal CA1 pyramidal neurons (‘n’ denotes the number of recorded neurons; Ctrl, n = 15; LATeNT, n = 17; **p < 0.01, ***p < 0.001, Mann-Whitney U test). For (B), p = 0.2903/0.0002/0.0002/0.0011/0.0018/0.0024 for 10/20/30/40/50/60 μA IPSC amplitude. For (C), p = 0.001. (D) Representative dendritic PPR traces and average PPR. Data are presented as mean ± SEM (Ctrl, n = 13; LATeNT, n = 14-15; **p < 0.01, ***p < 0.001, Mann-Whitney U test). p = 0.3389/0.0015/0.0007/0.616 for 50/100/200/500 ms ISI. (E) Same as G, with stimulating electrode at SP layer (somatic stimulation). (F, G) Representative traces for somatic eIPSCs, average eIPSC I-O curve (F) and average eIPSC I-O slope (G) from hippocampal CA1 pyramidal neurons (Ctrl, n = 12; LATeNT, n = 10; n.s., not significant. Mann-Whitney U test). For (F), p = 0.203/0.1802/0.2829/0.4176/0.4562/0.5943 for 10/20/30/40/50/60 μA IPSC amplitude. For (G), p = 0.0804. (H) Representative somatic PPR traces and average PPR. Data are presented as mean ± SEM (Ctrl, n = 11-12; LATeNT, n = 9; Mann-Whitney U test). p = 0.7544/0.3824/0.0674/0.3312 for 50/100/200/500 ms ISI.

Source data

Light-driven ion channels and pumps such as halorhodopsin (eNpHR3.0³³) have been widely used for causal neuroscience, due to their ability to control neural spiking on a millisecond timescale. To compare eNpHR directly to LATeNT, we repeated the in vivo manipulation of SST⁺ interneurons using eNpHR. In contrast to LATeNT-manipulated mice, eNpHR-mediated inhibition did not alter the time that mice spent in open arms in the EPM test (Extended Data Fig. 7). Thus, the anxiety phenotype we observed (Fig. 4e–j) requires the more sustained inhibition that LATeNT provides, highlighting an advantage of our tool over others.

Extended Data Fig. 7. Halorhodopsin (eNpHR)-mediated silencing of SST+ interneurons does not lead to changes in anxiety-like behaviors.

(A) Experimental procedure with eNpHR. Three weeks after AAV injection into CA1 of adult Sst-IRES-Cre mice, one cycle of EPM tests consisting of 3-min epochs with alternating laser manipulation (OFF-ON-OFF) was performed for each mouse. (B) Cre-dependent recombination of AAV-delivered genes. (C) Representative image showing eNpHR expression in hippocampal CA1 SST interneurons. Scale bar = 20 μm. (D) Heatmaps representing time spent in each arm of the EPM for mice of each group during each epoch (eNpHR group n = 7, EYFP group n = 7). (E) Same experiment as Fig. 4i except the halorhodopsin eNpHR3.0³⁴ was used to inhibit neuronal activity instead of LATeNT. Bar graphs show the time spent in open arms across trials for eNpHR-expressing mice versus non-expressing controls. Data are presented as mean ± SEM (n = 7 mice; n.s., not significant; Mann-Whitney U test). p = 0.9015/0.3829/0.8712 for OFF/ON/OFF. (F) Before and after graphs presenting the time spent in open arms across trials. Each pair of line-connected dots represents an individual mouse. 7 mice were analyzed per condition. (n.s., not significant; Friedman test). p = 0.1841 (control OFF vs ON), >0.9999 (control ON vs OFF), >0.9999 (eNpHR OFF vs ON), >0.9999 (eNpHR ON vs OFF). (G, H) Representative traces (G) and summary graphs (H) showing mean firing frequency (percentage) in eNpHR-EYFP-expressing SST+ neurons induced by current injected before and during yellow-light illumination (594 nm, 20 mW/cm²). Data are presented as means ± SEMs (n = 7–9 from 5 mice per group; ***p < 0.001; n.s., not significant; Friedman test). p = 0.0975 (control OFF vs ON), 0.5443 (control ON vs OFF), 0.0002 (eNpHR OFF vs ON).

Source data

NanoLuc–LATeNT for drug-activated VAMP2 cleavage in vivo

Although optogenetic tools provide high spatiotemporal precision, blue-light delivery to the mouse brain requires fiber implantation and is restricted to small regions. A drug-controlled version of TeNT could be easier to use across multiple brain regions at once. We fused LATeNT to the luciferase NanoLuc, which emits 460 nm blue light in the presence of its small-molecule substrate fluorofurimazine (FFz) (Extended Data Fig. 8a). Bioluminescence resonance energy transfer (BRET) from NanoLuc to LATeNT turns on LATeNT in the absence of externally applied blue light¹⁷.

Extended Data Fig. 8. NanoLuc-LATeNT* for drug-activated VAMP2 cleavage in vivo.

(A) Fusion of LATeNT to the luciferase NanoLuc allows LATeNT to be activated by a small-molecule (fluorofurimazine, FFz) instead of light. HEK293T cells expressing NanoLuc-LATeNT were treated with blue light for 30 min or FFz for 60 min before cell lysis and Western blot analysis. This experiment was performed 3 times with similar results. (B) Experimental procedure for testing NanoLuc-LATeNT* uncaging by cephalofurimazine (CFz) in the mouse brain. AAVs expressing NanoLuc-LATeNT* were injected into the hippocampal CA3 regions of one hemisphere, while AAVs expressing EGFP were injected into the contralateral CA3 regions. In +light control, optical fibers were implanted into the hippocampal CA1 regions one week after AAV injection, and light was delivered three weeks after AAV injection. For +CFz condition, mice were injected intraperitoneally with 1.3 µmol of CFz three weeks after AAV injection. (C) Representative bioluminescence images of mice expressing NanoLuc-LATeNT* in the hippocampus and injected with CFz. Control mice do not express LATeNT*. Scale bar = 1 cm. (D) Representative images of hippocampal CA1 regions stained with anti-cleaved VAMP2 (green) antibody. Scale bar = 20 μm. (E) Summary graphs quantifying cVAMP2⁺ puncta density in (D). Data are presented as mean ± SEM (n = 15 brain sections from 3 mice per group; *p < 0.05, **p < 0.01; ANOVA with Tukey’s post hoc comparisons test). p = 0.0063 (Control+CFz vs NanoLuc-LATeNT*+CFz), 0.0145 (NanoLuc-LATeNT* vs NanoLuc-LATeNT*+CFz), 0.024 (NanoLuc-LATeNT* vs NanoLuc-LATeNT*+Light).

Source data

In HEK293T-VAMP2 reporter cells, NanoLuc–LATeNT produced similar VAMP2 cleavage after 60 min of FFz as 30 min of light treatment (Extended Data Fig. 8a). To test NanoLuc–LATeNT* in vivo (we used the slow reset version of LATeNT for a larger response), we expressed it in the CA3 region and injected cephalofurimazine (CFz)³⁴ intraperitoneally. CFz is a furimazine analog optimized for in vivo use (Extended Data Fig. 8b). Functional expression of NanoLuc–LATeNT* was confirmed by in vivo bioluminescence imaging (Extended Data Fig. 8c). Immunofluorescence staining detected cleaved VAMP2 in tissue sections from CFz-injected mice but not in controls lacking LATeNT* expression or CFz administration (Extended Data Fig. 8d–e). The increase in cleaved VAMP2 was comparable to that observed in mice subjected to light via an implanted optical fiber. Thus, NanoLuc–LATeNT* provides an alternative for drug-controlled rather than light-controlled cleavage of VAMP2 and synaptic inhibition in vivo.

Control of other VAMP2-dependent pathways, including endocannabinoid signaling and insulin secretion

Though the major function of VAMP2 is to mediate presynaptic vesicle fusion and neurotransmitter release, it also plays a role in other membrane fusion events^35,36. We tested LATeNT for conditional VAMP2 cleavage in three such settings.

First, we tested LATeNT at postsynaptic sites, instead of presynaptic termini, to modulate the exocytosis of GluA1-containing AMPA receptors^24,37,38. In dissociated hippocampal neurons sparsely expressing LATeNT, we observed that 30 min of light significantly reduced surface GluA1 puncta (Extended Data Fig. 9a,b). The effect was absent in samples lacking LATeNT expression (Extended Data Fig. 9d,e). In this experiment, dendritic spine density remained unchanged (Extended Data Fig. 9c,f), probably because such structural remodeling requires longer than the 30 min activation time period that we used.

Extended Data Fig. 9. Post-synaptic effects of LATeNT.

(A) Representative images of cultured hippocampal neurons co-transfected with V5-LATeNT and EGFP vector for dendritic visualization. Following 30 min of stimulation with light, neurons were fixed and stained with antibodies against surface GluA1 (sGluA1), EGFP, and V5. Cyan arrows indicate the dendritic spine, and yellow arrows indicate surface GluA1 expression in spines. (B, C) Quantification of surface GluA1-positive puncta localized specifically to dendritic spines (B, p = 0.0173) and dendritic spine density per μm (C, p = 0.2913). Data are presented as mean ± SEM (LATeNT+Dark, n = 12; LATeNT+Light, n = 12; *p < 0.05; Mann-Whitney U test). (D–F) Experiments analogous to (A–C) performed in neurons expressing EGFP alone (without LATeNT) to assess the effect of light stimulation per se. Shown are representative images (D) and quantification of spine-localized sGluA1 (E, p = 0.8403) and spine density (F, p = 0.6866). Data are mean ± SEM (EGFP+Dark, n = 13; EGFP+Light, n = 13; Mann–Whitney U test).

Source data

Second, we used LATeNT to probe the mechanism of depolarization-induced suppression of inhibition (DSI)³⁹. DSI is a brain-wide mechanism for short-term plasticity in which depolarization of postsynaptic pyramidal cells (PCs) induces release of endocannabinoids (eCBs) that signal retrogradely to presynaptic cholecystokinin-expressing basket cells (CCK BCs) to transiently suppress incoming inhibitory transmission⁴⁰ (Fig. 5a). A recent study suggested that eCB release may be SNARE-dependent, since expression of constitutively active TeNT in PCs abrogated DSI⁴¹. However, this approach required population-level comparisons across animals and disrupted SNARE proteins for days to weeks, potentially allowing compensatory changes to occur in the circuit.

Fig. 5. Control of eCB signaling and insulin secretion using LATeNT.

a, Schematic of DSI. In DSI, depolarization of postsynaptic PCs triggers calcium influx, which leads to secretion of eCBs. Released eCBs activate cannabinoid type-1 receptor (CB1R) on presynaptic CCK BCs, which inhibits neurotransmitter release at the inhibitory synapse. Whether eCB release is SNARE-dependent is currently unclear. b, Experimental schematic for testing LATeNT’s effect on DSI. AAVs were injected bilaterally into CA1 of Sncg-IRES2-FlpO mouse that expresses FlpO specifically in CCK BCs⁴⁰. ChRmine-oScarlet is expressed in CCK BCs (driven by FlpO) and LATeNT is expressed in PCs (driven by CaMKII promoter). c, Interleaved pulses of red light (635 nm, 50 ms) and blue light (390 nm, 1.25 s) were delivered at 0.5 Hz for 20–45 min to activate ChRmine and LATeNT, respectively. Brief PC depolarization (0 mV for 1 s) was given every 2 min to assess DSI. d, Top row: example traces of ChRmine-induced oIPSCs at baseline. Bottom row: example oIPSC traces showing absence of DSI after ≥20 min of blue-light-mediated LATeNT activation in the same cell. Red rectangles indicate when ChRmine was activated. e, oIPSC amplitudes following depolarization (t = 0) from cells receiving 45 min of blue-light activation (blue; n = 8 cells from 5 mice (3 male, 2 female)), overlaid with oIPSC amplitudes at baseline (black; n = 11 cells from 7 mice (2 male, 5 female)). Note that baseline cells show DSI (oIPSC decrease post-depolarization), which is absent in cells receiving blue light. f, Paired plot showing percent DSI at baseline versus after 20–45 min of blue-light activation. Each dot represents 1 cell (n = 11 cells from 7 mice (2 male, 5 female), paired two-sided t-test; P = 0.0076; **P < 0.01. In e and f, data are presented as mean ± s.e.m. of normalized oIPSC amplitudes across cells within each time bin. g–i, Same as d–f with non-LATeNT activating red light used instead of blue light. In h, red: n = 8 cells from 5 mice (1 male, 4 female); black: n = 8 cells from 6 mice (2 male, 4 female). In i, n = 8 cells from 6 mice (2 male, 4 female), paired two-sided t-test; P = 0.2284; j, LATeNT controls endogenous insulin secretion. Top: experimental schematic; switching from low to high glucose stimulates insulin secretion from MIN6 pancreatic beta cells. Bottom: fold increase in glucose-stimulated insulin secretion (GSIS), detected by insulin ELISA, with/without doxycycline (Dox) induction of LATeNT expression and blue-light exposure for 30 min before low glucose treatment. Data are mean ± s.d. (n = 3); **P < 0.01; unpaired two-sided Student’s t-test. k, Western blot of wild-type and LATeNT-expressing MIN6 cells. P = 0.63 (light only), 0.3879 (Dox only), 0.0071 (Dox + light). Sncg, gamma-synuclein; nEF, short EF1a promoter; Coff/Fon, Flp-activated, Cre-inactivated cassette.

Source data

LATeNT offers a more precise method to acutely modulate eCB release and determine the role of VAMP2-mediated exocytosis in DSI. We expressed LATeNT specifically in postsynaptic PCs and the red-shifted opsin ChRmine^42,43 in presynaptic CCK BCs (Fig. 5b). We showed that LATeNT can modulate DSI at the population level: LATeNT-expressing slices showed a significant reduction in DSI following >45 min of blue-light (470 nm) exposure before depolarization (Extended Data Fig. 10a,b). Slices lacking LATeNT expression showed no change in DSI despite prolonged illumination (Extended Data Fig. 10c).

Extended Data Fig. 10. LATeNT inhibits endocannabinoid signaling.

(A) Left: Experimental setup of how cells were pre-incubated in the dark or under 470 nm blue light, followed by DSI measurements after a > 20-minute delay. Right: LATeNT modulation of DSI at the population level in experiments indicated in schematic on the left. Example traces of ChRmine-induced oIPSCs are shown from a cell maintained in the dark (top row) or in blue light (bottom row). (B) Plot showing percent DSI from pre-incubation experiments, corresponding to experimental setup in (A) (dark: n = 9 cells from 3 mice (1 male, 2 female), blue light: n = 8 cells from 4 mice (1 male, 3 female) (p = 0.0079); **p < 0.01, Mann-Whitney U test). (C) Same as (B) but in slices lacking LATeNT. Only ChRmine was expressed in CCK BCs (dark: n = 6 cells from 3 mice (1 male, 2 female), blue light: n = 5 cells from 3 mice (2 male, 1 female) (p = >0.9999); ns, not significant, Mann-Whitney U test). (D) Within-cell LATeNT activation, corresponding to Fig. 5f. Paired plot showing percent DSI at baseline versus after exactly 20 min of blue activation conditions. Each dot represents one cell. n = 10 cells from 6 mice (2 male, 4 female) (p = 0.0127), paired t-test, *p ≤ 0.05. (E) Within-cell control light activation, corresponding to Fig. 5i. Paired plot showing percent DSI at baseline versus after exactly 20 min of control light activation conditions. Each dot represents one cell. n = 8 cells from 6 mice (2 male, 4 female) (p = 0.0834), paired t-test, ns, not significant. In (B)-(E), data are presented as mean ± SEM of normalized oIPSC amplitudes across cells within each time bin.

Source data

We then performed long-term recordings from individual PCs before and after LATeNT activation. During the initial 2 min of baseline recording, brief pulses of red light were delivered to evoke inhibitory postsynaptic currents in PCs via ChRmine. These red light pulses were then interrupted by a 1-s depolarizing pulse, delivered through the recording electrode, to evoke DSI (Fig. 5c) detected as a transient suppression of optically evoked inhibitory postsynaptic currents (oIPSCs) (Fig. 5d). Following baseline recording, we interleaved the red-light (635 nm) pulses with 1.25-s pulses of 390 nm light to activate LATeNT. We specifically chose 390 nm to minimize unintentional blue-light activation of ChRmine⁴². We observed that LATeNT activation reduced DSI (Fig. 5d–f), with 20 min of light exposure being sufficient to produce an effect (Extended Data Fig. 10d). To rule out the possibility of unintentional ChRmine activation altering DSI, we performed a control experiment replacing 390 nm light pulses with non-LATeNT-activating red light pulses (‘control light’). In contrast to the blue-light group, DSI remained intact in the control group throughout the duration of the recordings (Fig. 5g–i and Extended Data Fig. 10e). Altogether, our experiments overcame the previous limitations of constitutively active TeNT by allowing temporally controlled, within-neuron comparisons, and demonstrate that postsynaptic eCB release is a SNARE-dependent process necessary for DSI in vivo.

For our third demonstration, we applied LATeNT in non-neuronal cells, to control the secretion of endogenous insulin from pancreatic beta cells. This secretion is dysregulated in type 1 and 2 diabetes, obesity and polycystic ovary syndrome, motivating the need for approaches to control insulin release. We expressed LATeNT in the cytosol of pancreatic MIN6 cells, an established model of glucose-stimulated insulin secretion (GSIS)⁴⁴. Cells were first starved in low glucose for 1 h and then stimulated with high glucose for 1 h to induce GSIS (Fig. 5j). In cells lacking LATeNT or kept in the dark, glucose stimulation increased GSIS by ~2.3-fold (Fig. 5j). By contrast, GSIS was impaired when LATeNT-expressing cells were stimulated with light for 30 min before glucose starvation (Fig. 5j). Western blotting and confocal imaging confirmed LATeNT- and light-dependent cleavage of endogenous VAMP2 in these cells (Fig. 5k and Supplementary Fig. 3).

Altogether, our findings show that LATeNT can be used to modulate diverse VAMP2-dependent processes in both neuronal and non-neuronal cell types.

Synthetic biology applications of LATeNT

Proteases are increasingly used as building blocks in synthetic circuits, allowing faster computations than gene transcription-based circuits. For such applications, sequence-specific orthogonal proteases are essential. Due to LATeNT’s high sequence-specificity, we wondered if it could be used for the construction of synthetic circuits in cells lacking endogenous VAMP2, such as fibroblasts.

First, we engineered LATeNT to respond to different inputs. Calcium ion (Ca²⁺) is a ubiquitous second messenger in many signaling processes including immune cell activation, neuronal activation, apoptosis and muscle contraction. To engineer a Ca²⁺-responsive version of LATeNT, we fused LATeNT to CaBLAM⁴⁵, a Ca²⁺-dependent luciferase. In HEK293T-VAMP2 reporter cells, ‘Ca-LATeNT’ showed 8.2 ± 1.6-fold greater cleavage of VAMP2 in the presence of Ca²⁺ (at supra-physiological levels) and the luciferase substrate FFz compared with FFz alone (mean ± s.d.; Fig. 6a). VAMP2 cleavage after 60 min of Ca²⁺/FFz was comparable to 30 min of light, suggesting efficient BRET activation of Ca-LATeNT. We also tested Ca-LATeNT in cultured neurons, and while cleavage of endogenous VAMP2 was detectable, the expression of Ca-LATeNT was poor in these cells (Supplementary Fig. 4).

Fig. 6. LATeNT for the construction of synthetic circuits in cells lacking endogenous VAMP2.

a, Calcium activated LATeNT through fusion of LATeNT to the Ca²⁺-activated luciferase CaBLAM⁴⁵. FFz is used by CaBLAM to generate bioluminescence, which activates LATeNT through BRET. This experiment was performed three times in HEK293T cells with similar results. See Supplementary Fig. 6e for western blots of other biological replicates. b, GPCR-LATeNT* converts GPCR activity into LATeNT* turn-on. HEK293T cells expressing GPCR-NanoLuc and LATeNT*–arrestin were stimulated with agonist (0.2 μg ml⁻¹ CCL20 or 1 μM salvinorin B) and FFz for 60 min or light for 30 min. This experiment was performed three times (CCR6) or two times (KORD) with similar results. See Supplementary Fig. 6f for western blots of other biological replicates. c, Gal4-LATeNT* is a single-component, light-dependent transcription factor. Top: domain structure of Gal4-LATeNT*. VAMP2(2–94) is the cytosolic domain of VAMP2. Bottom: anti-HA stain shows cytosolic localization of the construct. This experiment was performed three times in HEK293T cells with similar results. Fields of view from other biological replicates in Supplementary Fig. 6g. d, RELEASE-LATeNT enables light-triggered secretion of protein cargoes. Top: domain structure of RELEASE-LATeNT. Bottom left: LATeNT activity separates the cargo from its fused ER retention motif, allowing the cargo to traffick to the cell surface. Bottom right: HEK293T cells expressing RELEASE-LATeNT were stimulated with 470 nm light for 30 min and released SEAP (alkaline phosphatase) reporter activity in the media was measured 4 h later. E233A mutation inactivates protease active site of LATeNT. Data are mean ± s.d. (n = 4); CCR6, CC motif chemokine receptor 6; KORD, κ-opioid receptor DREADD; TM, transmembrane domain.

Source data

Next, we engineered LATeNT to respond to G protein-coupled receptor (GPCR) activation. We fused LATeNT* to arrestin, which is recruited to the chemokine GPCR CCR6 upon activation by its peptide ligand CCL20. The luciferase NanoLuc was fused to the cytosolic tail of CCR6, such that receptor activation in the presence of FFz leads to BRET between NanoLuc and recruited LATeNT*–arrestin (Fig. 6b). We observed that CCL20 treatment led to a 5.5 ± 2.8-fold increase in VAMP2 cleavage compared with unstimulated HEK293T-VAMP2 reporter cells, and no cleavage was detected in the absence of FFz (mean ± s.d.; Fig. 6b). A similar effect was observed using a different GPCR, the kappa opioid receptor-based DREADD (KORD)⁴⁶, which gave an increase in VAMP2 cleavage upon treatment with its ligand salvinorin B (SalB) (Fig. 6b). These examples show that LATeNT can be gated by diverse inputs, including Ca²⁺ and receptor activity.

We then explored LATeNT’s ability to drive useful outputs, such as gene expression and reporter secretion. To test LATeNT-induced gene expression, we created a fusion of Gal4, VAMP2 and LATeNT* and co-expressed the construct in HEK293T cells with a UAS-mCherry reporter (Fig. 6c). The fusion protein is excluded from the nucleus in the dark state, due to its large size (112 kD). Upon light exposure, LATeNT* cleaves VAMP2 intramolecularly, releasing the 38 kD Gal4 fragment for nuclear entry and UAS-driven gene expression (Fig. 6c). Microscopy shows that mCherry expression was detected only in cells exposed to light but not in dark controls (Fig. 6c).

To generate a more immediate, non-transcriptional output, we harnessed the RELEASE⁴⁷ system, which uses protease activity to control protein secretion from the endoplasmic reticulum (ER). Our design uses the cytosolic domain of VAMP2 as the protease cleavage sequence and secreted alkaline phosphatase (SEAP) as the secreted cargo (Fig. 6d). Upon LATeNT activation, VAMP2 cleavage separates SEAP from a C-terminal ER retention motif, allowing SEAP to traffic through the trans-Golgi network into the extracellular medium. In cells co-expressing RELEASE reporter and cytosolic LATeNT, 4-h light exposure led to a 4.6-fold increase in SEAP activity in the media compared with dark controls (Fig. 6d). Mutation of LATeNT’s active site (E233A) abolished SEAP secretion in both light and dark conditions (Fig. 6d).

Finally, to test the generality of LATeNT’s design, we inserted hLOV into the analogous location of botulinum toxin serotype A (BoNT/A) protease, which has moderate sequence similarity (50%) but high structural similarity to TeNT⁴⁸ (Supplementary Fig. 5a). With minimal linker optimization, the resulting light-activated BoNT/A (LABoNT/A) showed increased cleavage of its natural substrate SNAP25 upon 30-min light exposure (Supplementary Fig. 5b).

Discussion

LATeNT provides an example, along with several recently reported tools^16,17,49,50, of how conformation-switching domains (LOV, CaM-M13 module and so on) can be used to regulate enzyme activity through engineered allostery. Both LATeNT and our previously developed LOV-Turbo¹⁷ exhibit outstanding dynamic range, with minimal dark-state activity and near-parental activity in the light state. LATeNT’s single-chain, allostery-based design also results in superior dynamic range and performance compared with a two-component photoactivatable BoNT tool¹³ (PA-BoNT; Supplementary Text 1).

LATeNT adds to the constellation of effective tools for synapse manipulation (Supplementary Table 1), with some notable advantages. Compared with light-induced cell ablation⁵¹, light-induced clustering of synaptic vesicles^52,53, constitutively active TeNT protease⁷ or overexpression of inwardly rectifying potassium (Kir) channels^54,55, LATeNT is more mechanistically and temporally precise, circumventing non-specific toxicity and the compensatory changes in neuronal circuitry caused by long-term tool activity.

Drug- and light-controlled GPCRs such as DREADDs⁴⁶ and PdCO²⁷ are widely used for causal neuroscience. LATeNT may be more potent and provide more sustained inhibition of synaptic transmission. Furthermore, its effect is more specific, because LATeNT acts solely through VAMP2 cleavage whereas Gi-coupled GPCRs activate multiple downstream pathways (in addition to inhibiting synaptic vesicle release through suppression of presynaptic VGCCs). See Supplementary Text 4 and Supplementary Table 1 for a detailed discussion comparing LATeNT with other synapse inhibition tools.

Several lines of future work could improve the utility and scope of LATeNT. First, engineering LATeNT to cleave proteins other than VAMP2 could expand the space of possible applications. Discovering a peptide recognition sequence for LATeNT/TeNT (to replace VAMP2) would also be helpful for creating more compact tools or synthetic circuits. Second, given LATeNT’s promise in controlling endogenous insulin secretion, it is worth exploring LATeNT for other related applications—such as control of glucagon secretion from pancreatic alpha cells and sense-and-respond technologies that integrate wearable, real-time glucose sensors. Third, if LATeNT activity can be regulated by cell–cell contact, then this could provide a powerful way to manipulate the activity of defined synaptic connections in the brain. To assist future development, we provide a detailed discussion of unsuccessful engineering attempts in Supplementary Text 5.

Methods

Cloning

For LATeNT constructs, gene fragments were ordered from Twist Biosciences and were subcloned into a pcDNA3 backbone (https://www.addgene.org/vector-database/2093/) using HindIII and XhoI restriction sites. All constructs were generated using standard cloning techniques. PCR fragments were amplified using Q5 polymerase (NEB, catalog number M0491S). Vectors were digested using enzymatic restriction digest and ligated to gel purified PCR products using Gibson assembly. Ligated plasmid products were transformed into competent XL1-Blue Escherichia coli. Detailed amino acid sequences are given in Supplementary Text 6.

Mammalian cell culture and transfection

HEK293T cells from ATCC (<30 passages) were cultured as a monolayer in complete media (Dulbecco’s modified Eagle’s medium (DMEM; Gibco, catalog number 11965-092) supplemented with 10% (w/v) fetal bovine serum (FBS; VWR, catalog number 97068-085) and 1% penicillin–streptomycin (VWR, catalog number 16777-164)) at 37 °C under 5% CO₂. In all experiments, glass coverslips and plates were pretreated with 10 µg ml⁻¹ human fibronectin (Millipore, catalog number FC010) in Dulbecco’s PBS (DPBS; Gibco, catalog number 14190-144) for 30 min at room temperature before cell plating. For transient expression, cells were transfected at 50–70% confluency with indicated expression plasmids using PEI (Polysciences, catalog number 24765-1) in DMEM without FBS.

Generation of VAMP2-expessing HEK293T cells

Reporter cells expressing mCherry–myc–VAMP2 were generated by lentiviral transduction and antibiotic selection. For lentivirus generation, HEK293T cells were cultured in six-well plates and were transfected at approximately 70% confluency with 1,000 ng of the lentiviral vector of interest, packaging plasmids pCMV-dR8.91 (900 ng) and pCMV-VSV-G (100 ng) with 12 µl of PEI. Approximately 48 h after transfection, the cell medium was collected, filtered through a 0.45-µm filter and then stored in 200-µl aliquots (‘supernatant lentivirus’). The aliquots were flash-frozen in liquid nitrogen and stored at −80 °C. For lentiviral transduction, HEK293T cells (<10 passages) were cultured in six-well plates and transduced with 200 µl of supernatant lentivirus at approximately 50% confluency. A total of 2 days after transduction, cells with stably integrated reporters were selected by maintaining the cells in complete media containing antibiotics (250 µg ml⁻¹ hygromycin for mCherry–myc–VAMP2). Antibiotic selection was performed until all cells showed mCherry signal under a tabletop fluorescence microscope. After selection, reporter cell lines were grown in T25 tissue culture flask until confluency, and frozen stocks were prepared and stored in liquid nitrogen for future use.

Rat cortical neuron culture and AAV infection

All procedures were approved and carried out in compliance with the Stanford University Administrative Panel on Laboratory Animal Care, and all experiments were performed in accordance with relevant guidelines and regulations (protocol number APLAC-32980). Before plating, plates were coated with 0.001% (wt%/vol%) poly-L-ornithine (Sigma-Aldrich, catalog number P4957) in DPBS at room temperature overnight, washed twice with DPBS, and subsequently coated with 3.3 µg ml⁻¹ of mouse laminin (Gibco, catalog number 23017015) in DPBS at 37 °C for 2–4 h. After laminin coating, coverslips were washed twice with DPBS and stored at 4 °C until plating. Cortical neurons were extracted from embryonic day 18 Sprague Dawley rat embryos (Charles River Laboratories) by dissociation in Hank’s balanced salt solution (HBSS; Gibco, catalog number 14025076). Cortical tissue was digested in papain according to the manufacturer’s protocol (Worthington, catalog number LK003150), then plated onto 0.1-mm-thick glass coverslips in complete neurobasal medium (CNB) at 37 °C under 5% CO₂. CNB is neurobasal (Gibco, catalog number 21103049), supplemented with 2% (v/v) B27 supplement (Life Technologies, catalog number A3582801), 0.1% (v/v) FBS, 1% GlutaMAX (Gibco, catalog number 35050061), 1% penicillin–streptomycin and 1 mM sodium pyruvate (Gibco, catalog number 11360070). On DIV1, half of the media was removed from each well and replaced with equal volume of CNB. On DIV4, neurons were infected with purified AAVs along with a media change. Neuronal cultures were wrapped in aluminum foil and were allowed to express for an additional 7 days in the incubator until light stimulation and subsequent analysis.

Light stimulation of LATeNT in cultured mammalian cells and neurons

After cells were transfected with LATeNT plasmids or transduced with LATeNT AAVs, the plates were covered with aluminum foil until ready for blue-light stimulation. When uncovered, cells were handled in a dark room under red light to avoid undesired activation of LATeNT. To stimulate LATeNT, cells were incubated at 37 °C while being placed directly on top of a blue-light LED array. We used an AMUZA system consisting of a blue LED array, an LED Array Driver and pulse generator. We used a 16% duty cycle (2 s on and 10 s off) for 30 min, unless noted otherwise. Light power before pulse generation was typically around 1.0 mW cm⁻² (measured by Coherent Fieldmax II TO laser power meter), unless noted otherwise. Control samples were processed in parallel omitting light. Following light stimulation, cells were once again handled under red light, washed with DPBS three times and analyzed by western blot or immunofluorescence as described below.

Western blot detection of LATeNT activity using anti-cleaved VAMP2 antibody

Following light stimulation and DPBS washes, cells were lysed directly in the cell culture wells with RIPA lysis buffer supplemented with 1× protease inhibitor cocktail (PIC; Thermo Scientific, catalog number 78429) and 10 mM of picolinic acid (Sigma-Aldrich). Lysates were cleared via centrifugation at 20,000g at 4 °C for 10 min. Cleared lysates were mixed with protein loading buffer and boiled at 95 °C for 10 min. The concentrations of cell lysates were normalized using a Pierce BCA Protein Assay Kit (Pierce, catalog number 23225), and samples were loaded onto a polyacrylamide gel, and transferred onto PVDF membranes (Cytiva, catalog number 10600023). Blots were blocked in 2% (wt%/vol%) nonfat milk (Lab Scientific, M-0841) in 1× Tris-buffered saline with Tween (TBST) (Teknova, catalog number T1688PK) for 15 min at room temperature, incubated with mouse antibody recognizing the N-terminal fragment of cleaved VAMP2 (VAMP/B/1148)²⁶ for overnight at 4 °C in the blocking buffer. After primary antibody staining, the membrane was washed three times with TBST for 5 min each, incubated in anti-mouse-HRP secondary antibody for 90 min at room temperature in 2% (w/v) bovine serum albumin (BSA; Fischer Scientific, catalog number BP1600-1) in TBST. After secondary antibody staining, the membrane was washed three times with TBST, rinsed twice with dH₂O, and chemiluminescence was developed with Clarity Western ECL Substrate (Bio-Rad) and imaged using a ChemiDoc XRS (Bio-Rad) imaging system or FlourChem M imaging system (ProteinSimple). Except for the VAMP/B/1148 antibody, other primary antibody stains, including the antibody that recognizes the C-terminal fragment of cleaved VAMP2 (VAMP/B/151)²⁶ (used in Fig. 1d,e, and Supplementary Figs. 1b and 2a), were done in 2% BSA in TBST, either for 2 h at room temperature or for overnight at 4 °C.

Immunofluorescence detection of LATeNT activity using anti-cleaved VAMP2 antibody

For confocal fluorescence microscopy experiments, cells were grown on 12-mm-diameter glass coverslips (VWR, catalog number 72230-01) in 24-well plates. Cells were transfected with LATeNT expression plasmids and stimulated with blue light as described above. After light stimulation, cells were gently washed three times with DPBS and were fixed with 4% (v/v) paraformaldehyde (PFA; ChemCruz, catalog number sc-281692) at room temperature for 15 min. After fixation, PFA was aspirated and cells were permeabilized for 5 min with ice-cold methanol. Cells were washed again three times with DPBS and blocked for 1 h with 1% bovine serum albumin (BSA; Fischer Scientific, catalog number BP1600-1) in TBST at 4 °C. Cells were then incubated with primary antibodies in TBST overnight at 4 °C. In immunofluorescence detection of LATeNT activity, mouse antibody recognizing the C-terminal fragment of cleaved VAMP2 (VAMP/B/151)²⁶ was used, as VAMP/B/1148 staining was not detectable in fixed cells. After washing three times with TBST, cells were incubated with fluorophore-conjugated secondary antibodies in TBST for 1 h at room temperature. Cells were washed three times with TBST and imaged.

Imaging was performed with a Zeiss Axio Observer.Z1 microscope with a Yokogawa spinning disk confocal head, Cascade IIL:512 camera, a Quad-band notch dichroic mirror (405/488/568/647 nm) and 405 nm, 491 nm, 561 nm and 640 nm lasers (all 50 mW). Images were captured through a 63× oil-immersion objective for the following fluorophores: AlexaFluor 405 (405 laser excitation, 445/40 emission), EGFR and AlexaFluor 488 (491 laser excitation, 528/38 emission), mCherry and AlexaFluor 568 (561 laser excitation, 617/73 emission) and AlexaFluor 647 (647 laser excitation, 700/75 emission). Image acquisition times ranged from 10 to 500 ms per channel, and images were captured as the average of two or three such exposures in rapid succession. Image acquisition and processing was carried out with the SlideBook 5.0 software (Intelligent Imaging Innovations, 3i).

Construction of yeast strains

All strains were derived from S. cerevisiae strain BY4741. Plasmid transformation or integration in yeast was performed using the Frozen E-Z Yeast Transformation II kit (Zymo Research, catalog number T2001) according to the manufacturer’s protocol. S. cerevisiae strains were produced stepwise and propagated at 30 °C in complete minimal media (CSM) with 20 g l⁻¹ dextrose (CSM-D). CSM is 6.7 g l⁻¹ yeast nitrogen base without amino acids (Thermo Fischer Scientific, catalog number H26271) and 0.54 g l⁻¹ CSM-Ade-His-Leu-Lys-Trp-Ura (Sunrise Science Products, catalog number 1135-010).

Transformants were isolated in appropriate selective medium by auxotrophic complementation. For yeast strain transformation, we grew cells at 30 °C in YPD containing 10 g l⁻¹ yeast extract (Gibco, catalog number 212750), 20 g l⁻¹ peptone (Gibco, catalog number 211677) and 20 g l⁻¹ dextrose. We first obtained the yeast strain containing the membrane tethered transcription factor and the reporter gene, by integrating the plasmid containing lexO::YFP–URA and ACT1::LexA–VP16–VAMP2, along with a HIS3 gene. Transformed cells containing the desired integration were selected on CSM plates supplemented with 100 mg l⁻¹ leucine and 800 mg l⁻¹ uracil (CSM-D + Leu, Ura).

With lexO::YFP reporter cells in hand, we episomally introduced plasmids containing GAL1::TeNT–mCherry or GAL1::LATeNT1–mCherry, along with a LEU2 gene. Transformed yeast cells containing the plasmid were selected on CSM plates supplemented with 800 mg l⁻¹ uracil.

Yeast culture and analysis of LATeNT activity in yeast cytosol

Yeast strains containing the YFP reporter, VAMP2-tethered transcription factor (LexA–VP16–VAMP2), and galactose-inducible LATeNT–mCherry (or TeNT–mCherry) were propagated at 30 °C in CSM-D media supplemented with 800 mg l⁻¹ uracil (CSM-D + Ura). To induce protein expression, yeasts were inoculated from saturated cultures in CSM with 2 g l⁻¹ dextrose and 18 g l⁻¹ galactose (CSM-D/G) overnight from a 1:20 dilution. After inoculation, culture tubes were wrapped in aluminum foil to prevent light exposure.

After overnight induction, the culture was exposed to ambient room light for 1–10 min, wrapped in aluminum foil and cultured at 30 °C for additional 8 h to allow reporter expression. Cells were then collected from 1 ml of cell culture, by pelleting at 3,000g for 2 min at 4 °C and resuspending in 0.1 ml PBS with 0.1% BSA (PBS-B). To analyze and sort, single yeast cells were gated on a forward-scatter area (FSC-A) by side-scatter area (SSC-A) plot around the clustered population (P1) on a ZE5 Cell Analyzer (Bio-Rad). P1 was then gated on a FSC-A by forward-scatter height (FSC-H) around the clustered population (P2). P2 populated cells were then plotted to detect mCherry and YFP signals. FlowJo v10 (BD Biosciences) was used to analyze FACS data.

Library generation for directed evolution of LATeNT in yeast

Mutant LATeNT1 library was generated using error-prone PCR with 100 ng of plasmid encoding GAL1::mCherry–LATeNT1 as the template. Three libraries were generated with different levels of mutagenesis, by carrying out error-prone PCR with different levels of mutagenesis:

• E1-1 (low mutation): 5 μM 8-oxo-dGTP, 1 μM dPTP, 15 PCR cycles

• E1-2 (medium mutation): 10 μM 8-oxo-dGTP, 2 μM dPTP, 15 PCR cycles

• E1-3 (high mutation): 15 μM 8-oxo-dGTP, 4 μM dPTP, 15 PCR cycles

Error-prone PCR was carried out following published protocols⁵⁶, with following primers that restricted the mutation to the LATeNT1 region:

• F: GCCACCatgGCTAGCGTTAACccgatcacc

• R: ttgtcctcctcgcccttgctca

PCR products were gel purified then reamplified for 30 more cycles under normal conditions and gel purified again. The original plasmid was digested with NheI and MscI and gel purified as well, to serve as a backbone for DNA recombination. Both 4,000 ng of PCR product and 1,000 ng of cut vector were mixed and water was added to total 10 μl. The resulting mixture was electroporated into electrocompetent lexO::YFP reporter yeasts. After electroporation, cells were rescued in 2 ml of YPD media and recovered at 30 °C for 1 h. Then, 1.98 ml of the culture was propagated in 100 ml of CSM-D + Ura media, while the remaining 20 µl was used to determine library size. Yeasts were diluted 100×, 1,000×, 10,000×, 100,000× and 20 µl of each dilution was plated onto CSM-D + Ura plates at 30 °C for 3 days. The resulting library size was determined by the number of colonies on the 100×, 100×, 1,000× and 100,000× plates, corresponding to 10⁴, 10⁵, 10⁶ or 10⁷ transformants in the library, respectively. The library sizes resulting from the three libraries were 3.5 × 10⁷, 4.0 × 10⁷ and 4.0 × 10⁷, respectively. All three libraries were tested and showed expected light-dependent YFP expression compared with the original LATeNT1 strain. All libraries were combined with equal cell numbers to perform subsequent directed evolution.

Directed evolution of LATeNT in yeast

Yeast cells were sorted using Sony SH800 sorter (Sony Biotechnology). Gates for singlets were drawn using FSC-A, SSC-A and FSC-H values as previously described. The resulting singlet population was drawn on a mCherry by YFP plot to collect cells with desired activity and expression level.

In round 1, cells were induced while exposed to ambient light. We collected the top 1.03% of cells with high YFP/mCherry ratio, as these cells will contain LATeNT variant that is active under light.

In rounds 2 to 5, cells were induced while wrapped in aluminum foil. Two different selections were done alternatively, either sorting cells with high mCherry/YFP ratio after light stimulation (positive control) or sorting cells with low YFP/mCherry ratio without light stimulation (negative control). Each selection was performed to enrich LATeNT mutant with high light-dependent activity or with low background activity at dark. Rounds 2 and 4 were negative selection with top 2.2% and 2.6% of cells being collected, respectively. Rounds 3 and 5 were positive selection with top 1.7% and 0.4% of cells being collected, respectively.

After five rounds of enrichment, plasmids in the enriched library was extracted using the Zymoprep yeast Plasmid Miniprep II kit (Zymo Research, catalog number D2004) following manufacturer’s protocol. Then, plasmids were transformed into competent XL1-Blue E. coli, single colonies were grown overnight, and plasmid was extracted and sequenced by Sanger sequencing.

In vitro VAMP2 cleavage assay by TeNT and LATeNT

For in vitro VAMP2 cleavage assay, wild-type HEK293T cells were transfected with mCherry–myc–VAMP2, TeNT or LATeNT expression plasmid, as described above, and cells were kept in the dark. 24 h after transfection, cells were stimulated with light for 30 min and cytosolic proteins were extracted using M-PER mammalian protein extraction reagent (Thermo Scientific, catalog number 78501) under ambient room light. The soluble fraction of VAMP2 lysate and TeNT (or LATeNT) lysate were mixed and incubated under ambient room light. The reaction was quenched by adding 50 mM of picolinic acid to inhibit protease activity. The reaction mixture was further analyzed by western blotting.

BRET activation of NanoLuc–LATeNT, Ca-LATeNT or GPCR-LATeNT*

For LATeNT activation by BRET, cells were cultured and transfected with LATeNT constructs as described above. After 24 h of transfection, the media was replaced with a media containing 50 μM of FFz (Selleck Chemicals, catalog number E1620) in the dark room and incubated at 37 °C for 1 h. For Ca²⁺ stimulation, CaCl₂ and ionomycin were added to a final concentration of 5 mM and 2 μM, respectively. For GPCR activation, GPCR agonists CCL20 or salvinorin B was added to a final concentration of 0.2 μg ml⁻¹ or 1 μM, respectively. After FFz treatment, cells were washed with DPBS and analyzed by western blotting as described above.

Light-dependent reporter expression using Gal4-LATeNT*

For Gal4-LATeNT* experiments, HEK293T cells were cultured on coverslips and transfected with Gal4–VAMP2–LATeNT* constructs as described above. After 24 h of transfection, cells were stimulated with light for 30 min and kept in dark for additional 24 h to allow reporter expression. Cells were PFA-fixed in the dark room and subsequently stained with anti-HA antibody to visualize reporter expression. mCherry expression was quantified by confocal imaging.

RELEASE-LATeNT assay

For RELEASE-LATeNT, cells were cultured, transfected with LATeNT constructs and stimulated with light as described above. After light stimulation, cells were covered with aluminum foil and incubated at 37 °C for 4 h. SEAP activity in the cell culture supernatant was measured following published protocols⁴⁷.

GSIS assay

MIN6 cell culture and GSIS assay were performed following published protocols^44,57. In brief, cells were maintained in DMEM supplemented with 15% heat-inactivated FBS, 2 mM GlutaMAX, 1 mM sodium pyruvate, 1% penicillin–streptomycin and 0.05 mM beta-mercaptoethanol. MIN6 cells were transfected with doxycycline-inducible LATeNT using lentiviral delivery and selected by puromycin. For GSIS assay, cells were starved for 1 h under KRB buffer⁵⁷ containing 1 mM glucose. After starvation, media was replaced with KRB buffer with either low (1 mM) or high (25 mM) glucose concentration. After 1 h, cell culture supernatant was collected and remaining cells were lysed. The amount of insulin in the supernatant and the lysate was measured using insulin ELISA kit (Mercodia) following manufacturer’s protocols. For LATeNT-expressing MIN6 cells, cells were either kept in dark or stimulated for 30 min with ambient room light before starvation.

Animals

Sst-IRES-FlpO (028579, Jackson Research laboratories), Sst-IRES-Cre (013044, Jackson Research Laboratories) and C57BL/6J mice (purchased from Daehan Biolink) used in the current study were maintained and handled in accordance with protocols (DGIST-IACUC-20122401-0004) approved by the Institutional Animal Care and Use Committee (IACUC) of the Daegu Gyeongbuk Institute of Science and Technology (DGIST). Mice were maintained on a 12–12 h light–dark cycle under standard temperature (22 ± 2 °C)-controlled laboratory conditions and received water and food ad libitum. Pregnant rats purchased from Daehan Biolink were used for in vitro culture of dissociated hippocampal neurons. All experimental procedures were conducted according to guidelines and protocols for rodent experimentation approved by the IACUC of DGIST. For DSI experiments, Sncg-IRES2-FlpO mice (034424, Jackson Research Laboratories) were maintained on a 12–12 h light–dark cycle under standard temperature (70–76 °F) and humidity (30–60%) and handled in accordance with protocols approved by Stanford University Administrative Panel on Laboratory Animal Care (APLAC).

Electrophysiology recordings in cultured hippocampal neurons

Cultured hippocampal neurons were infected at DIV4 with AAVs expressing mCherry–LATeNT or EGFP–PdCO (Addgene plasmid #198513) and kept in the dark. At DIV14–16, neurons were either kept in the dark, stimulated with 470 nm LED light for 30 min or exposed to 395 nm UV light for 5 min, and subsequently analyzed using whole-cell patch-clamp electrophysiological recordings. Patch pipettes were pulled from borosilicate glass capillaries (outer diamter, 1.5 mm; inner diameter, 0.86 mm; Sutter Instrument) using a P-97 micropipette puller (Sutter Instrument). The resistance of patch pipettes filled with internal solution varied between 3 and 6 MΩ. For recordings of mEPSCs, the composition of the internal solution was 145 mM CsCl, 5 mM NaCl, 10 mM HEPES, 10 mM EGTA, 0.3 mM Na-GTP and 4 mM Mg-ATP, with pH adjusted to 7.2–7.4 with CsOH and an osmolarity of 290–295 mOsmol l⁻¹. For recording of eEPSCs, the composition of the internal solution was 130 mM Cs-MeSO₄, 5 mM TEA-Cl, 1 mM QX-314, 0.5 mM EGTA, 8 mM NaCl, 10 mM HEPES, 0.4 mM GTP-Na, 10 mM phosphocreatine-Na₂ and 4 mM ATP-Mg, with pH adjusted to 7.2–7.4 with CsOH and an osmolarity of 290–295 mOsmol l⁻¹. The external solution consisted of 130 mM NaCl, 4 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES and 10 mM D-glucose, with pH adjusted to 7.2–7.4 with NaOH and an osmolarity of 300–305 mOsmol l⁻¹. The whole-cell configuration was obtained at room temperature using μM-TSC manipulators (Sensapex). Electrophysiological data were acquired with a Multiclamp 700B amplifier (Axon Instruments), and pCLAMP software and digitized using an Axon DigiData 1550B data acquisition board (Axon Instruments). mEPSCs and eEPSCs were recorded at a holding potential of −70 mV. Synaptic currents were analyzed offline using Clampfit 10.8 software (Molecular Devices). For the recordings of mEPSCs, the external solution included 1 μM TTX and 50 μM picrotoxin to block the Na⁺ currents and GABA_A receptors, respectively. For the recordings of eEPSCs, 10 µM picrotoxin were included in the extracellular solution. For measuring evoked synaptic responses were stimulated (0.2-ms current pulses) using a bipolar concentric electrode positioned 200–300 mm in front of postsynaptic pyramidal neurons at intensities producing 40–50% of the maximal EPSCs amplitude.

Measurement of surface GluA1 levels in cultured hippocampal neurons

Cultured hippocampal neurons were prepared from E18 rat brains cultured on coverslips coated with poly-D-lysine and grown in Neurobasal medium supplemented with B-27 (Invitrogen), 0.5% FBS, 0.5 mM GlutaMax (Invitrogen) and sodium pyruvate (Invitrogen). For sparse expression of LATeNT in cultured neurons, hippocampal neurons were co-transfected with pCAG-V5-LATeNT and EGFP expression plasmid⁵⁸ using CalPhos Kit (Clontech) at DIV8 and immunostained at DIV14. For surface GluA1 staining, neurons were first fixed with 4% formaldehyde/4% sucrose in PBS, then incubated with anti-surface GluA1 primary antibody before permeabilization. After surface staining, neurons were permeabilized with 0.2% Triton X-100 in PBS and subsequently incubated with antibodies against intracellular markers (for example, EGFP and V5). Appropriate Cy3- and fluorescein isothiocyanate-conjugated secondary antibodies (Jackson ImmunoResearch) were used for detection. Images were acquired using a confocal microscope (LSM800, Carl Zeiss) with a 63× objective lenses; all image setting were kept constant. Z-stack images were converted to maximal projection and analyzed to obtain the size, intensity, and density of puncta immunoreactivities of marker proteins. Quantification was performed in a blinded manner using MetaMorph software (Molecular Devices).

Electrophysiology recordings in acute brain slices

Six-week-old C57BL/6J or Sst-IRES-FlpO mice were injected in the hippocampal CA3 region with AAV-mCherry-LATeNT, and transverse hippocampal slices were prepared 2 weeks later. After anesthesia with isoflurane, mice were decapitated and their brains were rapidly removed and placed in an ice-cold oxygenated (95% O₂ and 5% CO₂), low-Ca²⁺/high-Mg²⁺ solution containing 3.3 mM KCl, 1.3 mM NaH₂PO₄, 26 mM NaHCO₃, 11 mM D-glucose, 211 mM sucrose, 0.5 mM CaCl₂ and 10 mM MgCl₂. Hippocampal slices were cut using a vibratome (VT1000s, Leica) and transferred to a holding chamber containing oxygenated artificial cerebrospinal fluid (aCSF; 124 mM NaCl, 3.3 mM KCl, 1.3 mM NaH₂PO₄, 26 mM NaHCO₃, 11 mM D-glucose, 3 mM CaCl₂ and 1.5 mM MgCl₂). Slices were incubated at 30 °C for at least 60 min and used for recordings within 4 h after slicing. For slices expressing LATeNT, experiments were performed under dark conditions unless otherwise indicated. When LATeNT activation was required, slices were exposed to ambient room light (~0.225 mW cm⁻²) for 1–2 h before recordings. In cases where specific blue-light illumination was applied, the detailed stimulation parameters are provided in the corresponding figure legends. Whole-cell recordings were performed at 30–32 °C using a MultiClamp 700B amplifier and DigiData 1550B digitizer (Molecular Devices) controlled by pCLAMP software. Slices were continuously perfused with oxygenated aCSF throughout recordings. Patch pipettes (3–8 MΩ) were filled with an internal solution containing 130 mM Cs-methanesulfonate, 5 mM TEA-Cl, 8 mM NaCl, 0.5 mM EGTA, 10 mM HEPES, 4 mM Mg-ATP, 0.4 mM Na-GTP, 1 mM QX-314 and 10 mM disodium phosphocreatine (280–290 mOsm). Signals were sampled at 10 kHz and filtered at 4 kHz. Cells were excluded if series resistance exceeded 30 MΩ. For measurement of evoked EPSCs (eEPSCs), PTX (50 μM) was applied to block the GABA_A receptor, and a concentric bipolar electrode was placed in the stratum radiatum. eEPSCs were recorded at −70 mV (for AMPA-EPSCs) and +40 mV (for NMDA-EPSCs; 50 ms after stimulation). Input–output responses were obtained by eliciting EPSCs with a series of stimulation intensities (20–100 μA). PPRs were calculated by delivering two pulses at different intervals (50, 100, 200 and 500 ms). To measure changes in eEPSC kinetics, eEPSCs were evoked every 20 s under dark conditions for 5 min to establish a stable baseline, followed by 30 min of continuous illumination with 473 nm blue light (0.5 mW cm⁻²). Three responses obtained within each 1-min interval were averaged and plotted as a function of time. For measurement of evoked inhibitory postsynaptic currents (eIPSCs), AMPA and NMDA receptors were blocked with CNQX (10 μM) and D-AP5 (50 μM), respectively. A concentric bipolar stimulating electrode was placed either in the stratum pyramidale or stratum lacunosum-moleculare, and eIPSCs were recorded at 0 mV. Input–output curves were obtained by eliciting IPSCs with a series of stimulation intensities (10–60 μA). PPRs were calculated by delivering two pulses at different intervals (50, 100, 200 and 500 ms). For measurement of neurotransmission in long-range projections, mice were injected in the vHPP (AP, −3.5 mm; ML, ±3.5 mm; DV, −3.4 to −3.8 mm) with AAV-GFP-ChrimsonR together with AAV-mCherry-LATeNT and maintained for 6–7 weeks. To obtain coronal slices of the mPFC containing long-range projections from the vHPP, each whole brain was cut at a 10° angle relative to the coronal plane, beginning at the front of the forebrain. Whole-cell recordings were obtained from mPFC pyramidal neurons voltage-clamped at −70 mV using the internal solution described above. Optogenetically evoked EPSCs (oEPSCs) were elicited by red-light stimulation of ChrimsonR-expressing vHPP axon terminals (stimulation parameters specified in the corresponding figure legends). To isolate monosynaptic terminal-driven responses, recordings were performed sequentially under baseline (no drug) conditions, followed by bath application of TTX (1 μM) and, subsequently, TTX + 4-AP (100 μM). Light-evoked oEPSCs were abolished by TTX and reliably rescued by TTX + 4-AP. Quantification of oEPSC amplitudes was performed under TTX + 4-AP conditions. For eNpHR validation experiments, current-clamp recordings were performed using K-gluconate-based internal solution (130 mM K-gluconate, 20 mM KCl, 0.2 mM EGTA, 10 mM HEPES, 4 mM Mg-ATP, 0.3 mM Na-GTP and 10 mM disodium phosphocreatine, pH 7.2–7.3 adjusted with KOH). Action potentials were evoked in CA1 stratum oriens SST-positive interneurons by 300-pA current injections and compared in the absence and presence of yellow-light illumination. Experiment-specific light stimulation protocols are described in the corresponding figure legends.

Preparation and titration of AAVs

AAVs were prepared as described previously⁵⁹. In brief, HEK293T cells were co-transfected with pHelper and pAAV1.0, together with pAAV-CAG-mCherry-LATeNT, pAAV-EF1a-fDIO-mCherry-LATeNT, pAAV-CAG-NanoLuc-LATeNT*, pAAV-EF1a-DIO-eNpHR3.0-EYFP (Addgene #26966) or pAAV-EF1a-ChrimsonR-GFP (Addgene #122063). Cells were collected 72 h later, lysed, mixed with 40% polyethylene glycol and 2.5 M NaCl and centrifuged at 2,000g for 10 min. The resulting pellets were resuspended in HEPES buffer (20 mM HEPES, 115 mM NaCl, 1.2 mM CaCl₂, 1.2 mM MgCl₂ and 2.4 mM KH₂PO₄), mixed with an equal volume of chloroform and centrifuged at 400g for 7 min. The supernatant fractions were concentrated with a Centriprep centrifugal filter (Millipore 4310) at 2,500 rpm for 15 min each, followed by concentration with an Amicon Ultra centrifugal filter (0.5 ml) at 14,000 rpm for 10 min. Virus infection titers were determined by qRT–PCR detection of mCherry sequences based on a standard curve generated using the corresponding DNA plasmid.

Stereotaxic surgery and virus injections

Six-week-old C57BL/6J, Sst-IRES-FlpO or Sst-IRES-Cre mice, with heads firmly secured in a stereotactic device, were anesthetized by inhalation of isoflurane (2–3%). Virus solutions were injected using a Hamilton syringe at a flow rate of 0.1 µl min⁻¹. The indicated coordinates were used for stereotaxic injection into the hippocampal CA1 (AP, −2.5 mm; ML, ±1.5 mm, DV, −1.3 mm), hippocampal CA3 (AP, −2.1 mm; ML, ±2.3 mm; DV, −2.4 mm) or ventral hippocampal CA1 (AP, −3.5 mm; ML, ± 3.0 mm; DV, −3.4 mm). After 2- to 5-week recovery period, the injected regions were subjected to ex vivo electrophysiological analyses. For in vivo experiments, optical fibers (FC-ZF1.25 F; Doric Lenses) were implanted perpendicularly into the targeted brain region 1 week after viral injection. Mice with off-target viral infection, identified by post hoc histological analysis, were excluded from all analyses.

Enriched environmental protocol

Three weeks after viral injection, blue light (473 nm, 1 mW) was delivered to the dSub region for 30 min (2 s on/10 s off), after which mice were exposed to EE conditions for 8 h per day for five consecutive days, then subjected to functional analyses. Blue laser (BL473T3-100FC; Shanghai Laser and Optics Century) was controlled with an isolated pulse stimulator (MODEL 2100; A-M SYSTEMS) connected to the optic fiber (FC-ZF1.25 F; Doric Lenses). The EE consisted of a large cage containing a running wheel, hut, tunnel and several other novel objects, as previously described³⁰.

Mouse behavioral tests

EPM test

The elevated plus maze is a plus-shaped, white, acrylic maze with two open arms (30 × 5 × 0.5 cm³) and two closed arms (30 × 5 × 30 cm³) positioned at a height of 75 cm from the floor. Light conditions around open and closed arms were ~300 and ~30 lux, respectively. For the test, mice were placed into the center zone of the elevated plus maze and allowed to move freely for 5 min. For LATeNT-mediated inhibition, blue light (473 nm, 1 mW) was delivered to the hippocampal CA1 region for 30 min (2 s on/10 s off) before mice were tested in the EPM. For NpHR-mediated inhibition, continuous yellow light (594 nm, 5 mW) was delivered to the hippocampal CA1 region for 3 min while mice were tested in the EPM. All behaviors were recorded by a top-view infrared camera, and the time spent in each arm was measured and analyzed using EthoVision XT 10.5 software (Noldus Information Technology).

Open-field test

The open-field test was performed to assess locomotor activity and anxiety-like behavior. The apparatus consisted of a square arena measuring 35 × 35 cm with opaque walls 40 cm in height. Mice were transported to the testing room in their home cages and allowed to acclimate for at least 30 min. Each trial began with the mouse placed in a designated corner of the arena facing the wall, and behavior was recorded for a total of 30 min. The arena was cleaned with 70% ethanol and dried between trials to remove olfactory cues. Behavioral parameters included total distance traveled, time spent in the center zone and number of center entries. Data were analyzed using EthoVision XT.

Light/dark box test

The light/dark box test was performed to assess anxiety-like behavior. One compartment was brightly illuminated (~1,000 lux) and constructed of white opaque walls (light chamber), while the other was covered and painted black to provide a dark environment (<5 lux; dark chamber). Each mouse was placed in the center of the light compartment at the start of the trial and allowed to move freely between compartments for 7 min. Behavioral parameters (time spent in each compartment, number of transitions) were analyzed using EthoVision XT software.

Sucrose preference test

The sucrose preference test was used to assess anhedonia-like behavior. The procedure consisted of a 48-h habituation phase followed by a 24-h test session. During habituation, each mouse was housed individually and provided with two identical drinking bottles, both containing tap water, to minimize neophobia and to allow adaptation to the two-bottle choice setup. For the test session, one bottle was filled with tap water and the other with a 2% (w/v) sucrose solution. Bottles were identical in size and shape and were placed in symmetrical positions on the cage to prevent side bias. The positions of the sucrose and water bottles were switched after 12 h during the test to control for side preference.

Bioluminescence imaging of mice

A total of 3 weeks after injecting AAV-NanoLuc-LATeNT* into the hippocampal CA3 regions, mice were administered intraperitoneal injections of reconstituted CFz (1.3 µM), as previously described³⁴. Bioluminescence images were acquired using the IVIS Spectrum In Vivo Imaging System (Caliper Life Sciences). During imaging, mice were anesthetized with isoflurane delivered via an XGI-8 Gas Anesthesia System (Caliper Life Sciences). Image settings were as follows: emission filter open, field of view of 14 cm, f-stop of 1.0, height of 2.5 cm and exposure time of 1 s.

Immunohistochemistry and imaging

Mice were anesthetized and immediately perfused, first with PBS for 3 min and then with 4% paraformaldehyde for 5 min. Brains were dissected out, fixed in 4% paraformaldehyde overnight and then incubated with 30% sucrose (in PBS) overnight and sliced into 40-μm-thick coronal sections using a vibratome (Model VT1200S; Leica Biosystems). Sections were permeabilized by incubating with 0.2% Triton X-100 in PBS containing 5% bovine serum albumin and 5% horse serum for 1 h. For immunostaining, sections were incubated for 8–12 h at 4 °C with primary antibodies diluted in the same blocking solution. The following primary antibodies were used: anti-c-Fos (Cell Signaling, catalog number 2250, RRID: AB_2247211, 1: 500), anti-SST (Millipore, catalog number MAB354, RRID: AB_2255365, 1:50) or anti-cleaved VAMP2 antibodies (VAMP/B/151, detects C-terminal fragment). Sections were washed three times in PBS and incubated with appropriate Cy3- or fluorescein isothiocyanate-conjugated secondary antibodies (Jackson ImmunoResearch) for 2 h at room temperature. After three washes with PBS, sections were mounted onto glass slides (Superfrost Plus; Fisher Scientific) with Vectashield mounting medium (H-1200; Vector Laboratories). Z-stack images (five images, 3-μm thickness) were acquired in standard mode with a laser-scanning confocal microscope (LSM800 with Airyscan mode; Zeiss) and processed using the maximum intensity projection function in Zen2.6 software (Zeiss).

DSI recordings in acute brain slices

Two- to six-month-old Sncg-IRES2-FlpO mice (Sncg-Flp⁶⁰) were co-injected bilaterally in the dorsal hippocampal CA1 with AAV1/2-CaMKII-V5-LATeNT and AAV8-nEF-Coff/Fon-ChRmine-oScarlet (400 nl each per hemisphere, full titer). To maximize viral expression, two locations were injected in each hemisphere: 2.10 mm and 2.40 mm posterior, 1.50 mm medial and 1.35 mm ventral to bregma. After a minimum of 2 weeks to enable virus expression, mice were deeply anesthetized by Ketamine/Xylazine and isoflurane, and then transcardially perfused with an ice-cold protective recovery solution containing: 92 mM NMDG, 26 mM NaHCO₃, 25 mM glucose, 20 mM HEPES, 10 mM MgSO₄, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 2.5 mM KCl, 2 mM thiourea, 1.25 mM NaH₂PO₄, 0.5 mM CaCl₂, titrated to a pH of 7.3–7.4 with HCl. In a dark room (0.1–0.2 µW ambient light) minimally lit only by red light (620 nm), coronal slices (250 μm) containing the hippocampus were cut in ice-cold protective recovery solution using a vibratome (VT 1200 S, Leica Biosystems). Brain slices were then incubated in 35 °C protective recovery solution for approximately 12 min. Hippocampal slices were then kept in the dark (without red light) and maintained in room temperature aCSF consisting of: 126 mM NaCl, 26 mM NaHCO₃, 10 mM glucose, 2.5 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 1.25 mM NaH₂PO₄. All solutions were equilibrated with 95% O₂/5% CO₂.

All intracellular recordings were conducted in a dark enclosure (0.2 µW ambient light) illuminated with red light (>620 nm) only when positioning samples and equipment. Brain slices were submerged in a chamber perfused with oxygenated aCSF at 2.5 ml min⁻¹ and maintained at 33 °C by a chamber heater (BadController V, Luigs and Neumann). Hippocampal neurons were visualized using 775 nm IR-DIC illumination on an Olympus BX61WI microscope (Olympus Microscopy) with an sCMOS camera (Flash 4.0 LT+, Hamamatsu). Recording pipettes were pulled from thin-walled borosilicate capillary glass (King Precision Glass) using a P97 puller (Sutter Instrument) and were filled with: 130 mM KCl, 10 mM HEPES, 2 mM MgCl₂, CaCl₂, 1 mM EGTA, 2 mM ATP-Mg (pH-adjusted to 7.3 with KOH, osmolarity 290 mOsm). Pipettes had a 3–5 MΩ tip resistance. Glutamate receptor blockers were included in the bath to isolate inhibitory currents: 20 µM D-APV (Cayman Chemical, #14539) and 5 µM NBQX (Tocris, #1044).

Whole-cell voltage clamp recordings were performed on PCs in the dorsal CA1 of the hippocampus. Neurons were held at −65 mV and optical illumination was provided by a DG-4 arc lamp (Sutter instrument) to activate ChRmine (635 nm, red light, 100 µW) or LATeNT (390 nm, blue light, 170 µW). Initially, a 2-min baseline recording was conducted in which 50-ms pulses of red light were delivered at 0.5 Hz to activate ChRmine and interrupted by a 1-s depolarization pulse (0 mV) to evoke DSI. ChRmine activating pulses were then interleaved with 1.25-s pulses of red light (control, non-LATeNT-activating condition) or blue light (experimental, LATeNT-activating condition) at 0.5 Hz for 20–45 min. To assess whether prolonged blue-light activation itself affected eCB release properties, we performed a cross-neuron comparison using slices with LATeNT that were pre-incubated in the dark or under a blue-light flashlight (400 µW) for at least 45 min. Neurons from these slices were then tested for DSI using the same baseline protocol described above. To evaluate that blue light alone does not affect DSI, we also repeated the light pre-incubation experiments in slices without LATeNT expression. oIPSCs amplitudes were measured as the maximum inward current in the window between start of the optical stimulation to 100 ms after the end of the optical stimulation. Successful oIPSCs were defined as having an amplitude >3 s.d. beyond the mean current in the 100 ms before ChRmine stimulation. Neurons with a success rate <60% during 30 initial ChRmine stimulations were not included in DSI measurements. Optically eIPSC amplitudes were normalized to the average response of each cell to ChRmine stimulation before depolarization. Percent change in DSI was calculated as

$$\text{DSI}\left(\%\right)=100\times \left(1-\frac{\text{mean post-depolarization oeIPSC amplitude}}{\text{mean pre-depolarization oeIPSC amplitude}}\right).$$

Data were acquired in pClamp software (Molecular Devices) using a Multiclamp 700B amplifier (Molecular Devices), low-pass filtered at 2 kHz and digitized at 10 kHz (Digidata 1440 A, Molecular Devices). Data analysis was performed using Clampfit (Molecular Devices) and custom written Python scripts.

Statistical analyses

Data were assessed by one-way analysis of variance (ANOVA) with Tukey’s post hoc comparisons or Mann–Whitney U test, or Friedman test; ‘n’ numbers and tests used to determine statistical significance are stated in Figs. 1–6. The normality of data distributions was evaluated using the Shapiro–Wilk test. Origin 2022b (OriginLab) and Prism 10 (GraphPad Software) was used for the analysis of data and preparation of bar graphs. P values <0.05 were considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41592-026-03176-w.

Author contributions

H.R. and A.Y.T. conceived the project. H.R., J.W.U. and A.Y.T. designed experiments and analyzed the data. H.R. performed all LATeNT engineering, characterization, synthetic biology applications and endogenous VAMP2 cleavage assays in neurons, with the assistance of C.L. D.K. and H.K. performed stereotactic injections, optic fiber implantations, immunohistochemical analyses and behavioral analyses in mice. B.K., Y.J. and Y.K. performed electrophysiological analyses. S.M. and P.M.K. designed and performed DSI experiments with supervision by I.S. M.J. provided preliminary data on LOV insertion sites. F.X. performed LABoNT/A experiments. H.R., J.W.U. and A.Y.T. wrote the paper with input from all authors.

Peer review

Peer review information

Nature Methods thanks Mathias Mahn, J. Simon Wiegert and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Nina Vogt, in collaboration with the Nature Methods team.

Funding

This work was supported by the NIH (grant nos. R01MH135934 to A.Y.T. and I.S.; NS131728 to I.S.), the Wu Tsai Neurosciences Institute (Phil and Penny Knight Initiative for Brain Resilience to A.Y.T.), NSF (Neuronex award to A.Y.T.) and the Biohub, San Francisco. In addition, we are grateful for support from the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (grant nos. RS-2023-NR076948 to J.W.U.; RS-2026-25480152 to D.K.; and RS-2024-00339642 to H.K.). S.M. was supported by the Knight-Hennessy Scholars Program and the American Epilepsy Society Predoctoral Research Fellowship. A.Y.T. is a Biohub, San Francisco investigator.

Data availability

Key constructs are available via Addgene at https://www.addgene.org/browse/article/28259396/. All other data are available from the corresponding authors upon reasonable request. Source data are provided with this paper.

Code availability

The code used for DSI analyses is published in ref. ⁶⁰.

Competing interests

The authors declare no competing interests.

Extended data

is available for this paper at 10.1038/s41592-026-03176-w.

Supplementary information

The online version contains supplementary material available at 10.1038/s41592-026-03176-w.