Microscope setup
The complete optomechanical configuration is detailed in Extended Data Figs. 1 and 2 and Supplementary Video 1. The microscope was constructed on an inverted frame (ECLIPSE Ti2-E, Nikon). Four excitation lasers, 405 nm (OBIS LX 100 mW, Coherent), 488 nm (OBIS LS 150 mW, Coherent), 552 nm (2RU-VFL-P-2000-552, MPB Communications) and 642 nm (2RU-VFL-P-2000-642, MPB Communications), were combined using dichroic mirrors (DM1, LM01-613-25; DM2, LM01-503-25; DM3, LM01-427-25; all from Semrock) and modulated by an acousto-optical tunable filter (AOTF) (AOTFnC-400.650-TN, AA Opto-Electronic). The combined beam was coupled into a single-mode fibre (P1-488PM-FC-2, Thorlabs) by a coupling lens (PAF2P-A10A, Thorlabs). The fibre output was collimated by lens L1 (f = 50 mm; AC254-050-A, Thorlabs), expanded by lenses L2 (f = 100 mm; AC254-100-A, Thorlabs) and L3 (f = 250 mm; ACT508-250-A, Thorlabs) and focused onto the back focal plane of the objective by lens L4 (f = 400 mm; ACT508-400-A, Thorlabs) for sample illumination. An oil-immersion objective (UPLAPO100XOHR, ×100/1.5 NA, Olympus) was used for imaging structures near the coverslip, whereas a silicone oil-immersion objective (UPLSAPO100XS, ×100/1.35 NA, Olympus) was used for whole-cell and tissue imaging. Sample positioning was controlled by an automated stage (PZ-2000FT, Applied Scientific Instrumentation), and precise axial positioning of the objective was driven by a PIFOC piezo scanner (P-721.12Q, Physik Instrumente). After traversing the objective and sample, the excitation beam was retroreflected back by a protected silver mirror, generating the axial standing-wave interference pattern at the sample plane.
Fluorescence detection
Emitted fluorescence was separated from the excitation light using a dichroic mirror (DM4: ZT543rdc, Chroma) and filtered with a longpass emission filter (F1: ET560lp, Chroma). For single-colour imaging, signals exited through the right port of the microscope, passed through a cylindrical lens (CL1: f = 1000 mm; LJ1516RM-A, Thorlabs) to induce slight astigmatism for coarse axial localization and were focused onto an sCMOS camera (CAM1: ORCA-Fusion; C14440-20UP, Hamamatsu) with an additional emission filter (F2: ET560lp, Chroma (for cell imaging); ET605/70 m, Chroma (for tissue imaging)) placed in front of the camera. For two-colour imaging, signals exited through the left port, passed through a cylindrical lens (CL2: f = 1,000 mm; LJ1516RM-A, Thorlabs) and were relayed by a lens pair (L5 and L6/L7: f = 200 mm; 49364, Edmund). A dichroic mirror (DM5: ZT561rdc, Chroma) spectrally separated the light: conventional fluorescence passed through a bandpass filter (F3: ET605/70 m, Chroma), while salvaged fluorescence passed through a separate filter (F4: FF01-572/28-25, Semrock). A right-angle prism mirror (RAP120-RA-A, LBTEK) projected both channels onto distinct regions of a second sCMOS camera (CAM2: ORCA-Fusion; C14440-20UP, Hamamatsu). For dSTORM imaging of AF647, an alternative filter set was used: DM4 (ZT405/488/561/640rpcv2, Chroma), F1 (ZET405/488/561/640mv2, Chroma) and F2 (ET700/75m, Chroma).
Axial interference illumination module
The module design is illustrated in Extended Data Fig. 2 and Supplementary Video 1. To ensure a stable axial standing-wave interference pattern, the module was mechanically isolated from the microscope frame using a custom mount secured directly to the optical platform, positioned above the sample stage for optimal alignment. A 10 mm diameter protected silver mirror (M11; 34-386, Edmund) was attached to a piezoelectric actuator (S23.Z10K, CoreMorrow) for precise axial positioning. This assembly was mounted on a kinematic mirror mount (MT-AM1, LBTEK) to allow for coarse tip and tilt adjustments. A piezo linear stage (N-565.260, Physik Instrumente) provided coarse axial positioning and sufficient clearance for sample replacement.
Data acquisition
Two synchronization control modes were implemented. Slow imaging mode (Extended Data Fig. 3a) was driven by an NI-DAQ card (USB-6363, BNC, National Instruments) and the LabVIEW 2022 DAQ module. A periodic voltage waveform consisting of three 15 ms phases was generated by an analogue output port to control the piezoelectric actuator. Simultaneously, the sCMOS camera operated in external edge trigger mode, initiated by the rising edge of a digital output signal. To provide sufficient mechanical settling time for the mirror, the digital output signal was delayed by 3 ms relative to the analogue output signal. The camera exposure time was set to 10 ms, and the AOTF was maintained in continuous transmission mode throughout image acquisition. For fast imaging mode (Extended Data Fig. 3b), the sCMOS camera operated in internal trigger mode. A field-programmable gate array (PXIe-7867, National Instruments) monitored the camera readout signal and transmitted synchronized control signals to the AOTF to dictate the effective exposure time, as well as to the piezoelectric actuator to drive cyclic motion to predefined axial positions.
System calibration
To generate the calibration file for astigmatic localization, a sparse sample of 40 nm fluorescent beads (F8793, 580/605 nm, Invitrogen) was axially scanned over a range of 1 µm using the piezo stage at three distinct phases to acquire a 3D image stack. PSF widths were calculated to estimate the coarse z positions, revealing a linear relationship with the actual axial displacement. To calibrate the interference fringe period, the intensities of each phase were measured, and the positional and phase data were then fitted to determine the axial period of the interference fringes (Extended Data Fig. 4c). To determine the voltage of the piezoelectric actuator corresponding a 2π/3 shift in the interference fringe, a sparse sample of 40 nm beads was imaged while the sample remained stationary. The piezoelectric actuator was actuated using a 4–6 V analogue output signal. Frame-by-frame intensity measurements were fitted to a sine function to determine the oscillation period (2π), and the voltage corresponding to a 2π/3 shift was identified (Supplementary Fig. 2).
Reconstruction
The algorithm workflow is illustrated in Extended Data Fig. 5. We adopted previously developed pipelines9,12 with minor modifications. Sub-images (Z1, Z2, Z3) corresponding to the three phase-shifted illumination patterns were summed. Two-dimensional Gaussian fitting was used to estimate the lateral (x–y) positions and PSF widths (σx, σy) for each molecule. For 3D-SMLM, σx and σy were used to calculate coarse axial positions. For me4Pi-SMLM, the intensities of the sub-images were analysed to determine the phase and modulation depth along the z direction (Supplementary Note 1), and a ridge-finding algorithm was applied to unwrap the phases and determine precise z positions9 (Supplementary Fig. 1 and Supplementary Note 1). Colour assignment followed established protocols11. All images and videos were rendered using Vutara SRX software (Bruker). For live-cell imaging, data processing followed the same pipeline. For image reconstruction, localizations acquired within a defined time window (10 s in this study) were combined to generate a super-resolved image. For single-molecule tracking, 2D detection, localization and tracking were performed using TrackMate53. Subsequently, the axial position of each molecule was estimated using the aforementioned method and integrated with the 2D data to generate 3D trajectories. Trajectories shorter than ten frames were excluded from subsequent mobility analysis.
Drift correction
Given that me4Pi-SMLM uses interference illumination to achieve precise axial localization, both positional and phase drift can affect the final localization accuracy. To correct for phase drift, we adopted a previously developed cross-correlation approach54. In brief, localized molecules were divided into temporal segments, typically corresponding to a time window of 1,000–3,000 frames. A series of 2D histogram images representing phase and normalized metric of the molecules was generated and Gaussian-blurred (Supplementary Note 1). Cross-correlation was then used to estimate phase drift between images, which was subsequently corrected by spline interpolation. This approach substantially reduces phase drift, thereby enabling robust phase unwrapping and the reliable conversion of phase data into axial coordinates (Supplementary Fig. 1). For 3D sample drift correction, the AIM (adaptive intersection maximization)55 and DME (drift at minimum entropy)56 algorithms were applied sequentially, during which any residual phase drift was also addressed.
Buffers
The following buffers were used for DNA-PAINT or dSTORM imaging:
-
Blocking buffer A: PBS (1× PBS; Gibco, 10010023), 3% BSA (Jackson ImmunoResearch, 001-000-162), 0.2% Triton-X100 (Sigma-Aldrich, T8787);
-
Blocking buffer B: PBS, 3% BSA, 0.25% Triton-X100;
-
Blocking buffer C: PBS, 3% BSA, 1% Triton-X100;
-
Dilution buffer A: PBS, 1% BSA, 0.2% Triton-X100;
-
Dilution buffer B: PBS, 1% BSA, 1% Triton-X100;
-
Wash buffer: PBS, 0.1% Triton-X100;
-
Buffer C: PBS, 1 mM EDTA (Invitrogen, 91222915), 500 mM NaCl (Sigma-Aldrich, S5886-5KG), 0.02% Tween-20 (Sigma-Aldrich, P1379-500ML), pH 7.4;
-
Buffer D: 5 mM TCEP (Sigma-Aldrich, C4706-2G) + PBS + 1 mM EDTA, pH 6.8;
-
Trolox (100×) (Sigma-Aldrich, 238813-1 G): 100 mg Trolox, 430 µl 100% methanol (Sigma-Aldrich, 322415-100 ML), 480 µl 1 M NaOH (Sigma-Aldrich, S5881-500G) in 3.2 ml water;
-
Buffer E: PBS, 500 mM NaCl, 20 mM Na2SO3 (Sigma-Aldrich, 71988-250 G), 1 mM Trolox, pH 7.4;
-
Storage buffer (50% glycerol (Sigma-Aldrich, V900122), PBS);
-
Cytoskeletal buffer (CB buffer: 10 mM MES (Sigma-Aldrich, M3671-50G), 150 mM NaCl, 5 mM MgCl2 (Sigma-Aldrich, M2393-500G), 5 mM EGTA (Sigma-Aldrich, 03777-10 g), 5 mM glucose (Sigma-Aldrich, G8270-1kg), pH 6.1);
-
Base buffer (44% glycerol, 50 mM Tris pH 8.0 (Thermo Fisher Scientific, 15568025), 10 mM NaCl, 10% glucose).
DNA-PAINT docking and imager sequences
HPLC-purified azide-labelled, dye-labelled and quencher-labelled oligonucleotides were obtained from Ningbo Karebay Biochem and GENEray Biotechnology. Three docking strands (L5R2, S4 and LS4) and their corresponding imager strands were used in this study. Specifically, L5R2 was paired with LR2; S4 with P10, P33 and P81; and LS4 with LP10 and LP71. To minimize non-specific binding, the ‘L-Oligo’ sequences (L5R2, LS4, LR2, LP10 and LP71) were synthesized using left-handed DNA57. S4 and LS4 were used as orthogonal pairs for two-colour imaging. The nucleotide sequences for all docking and imager strands used in this study are detailed below:
Oligonucleotide sequence (5′–3′)
Docking strands
L5R2: azide_ACCACCACCACCACCACCA
S4: CCTTCAACATTTCTTCTAC_azide
LS4: CCTTCAACATTTCTTCTAC_azide
Imager strands
LR2: TGGTGGT_Cy3B
P10: Cy3B_AGAAGTAATGTGGAA_BHQ2
P33: ATTO643_AGAAGTAATGTGGAA_BBQ650
P81: ATTO Rho11_AGAAGTAATGTGGAA_BHQ2
LP10: Cy3B_AGAAGTAATGTGGAA_BHQ2
LP71: AF568_AGAAGTAATGTGGAA_BHQ2
Conjugation of secondary antibodies with docking strands
Docking strands were conjugated to goat anti-mouse (Jackson ImmunoResearch, 115-005-146, lot no. 171417) and donkey anti-rabbit (Jackson ImmunoResearch, 711-005-152, lot no. 159871) secondary antibodies as previously described21. In brief, 250 µg of antibody was prepared using 100 kDa MWCO Amicon Ultra centrifugal filters (Sigma-Aldrich, UFC5100) and reacted with a 20-fold molar excess of DBCO-sulfo-NHS ester crosslinker (Sigma-Aldrich, 762040) overnight in the dark at 4 °C to label surface-exposed lysine residues. Unreacted crosslinker was subsequently removed using a ZEBA spin desalting column (7 kDa MWCO, Thermo Fisher Scientific, 89882), followed by three washes with PBS using the 100 kDa MWCO centrifugal filters. DBCO-functionalized antibodies were recovered by filter inversion and centrifugation. Next, copper-free click chemistry was used to attach the azide-modified DNA oligonucleotides (S4 and LS4). A 20-fold molar excess of the respective azide-DNA was added to the DBCO-conjugated antibodies, gently mixed and incubated in the dark at room temperature (20−25 °C) for 1 h. The resulting DNA-antibody conjugates (goat anti-mouse-S4 and donkey anti-rabbit-LS4) were purified from free DNA through five PBS washes using 100 kDa MWCO centrifugal filters. The final recovered conjugates were adjusted to a concentration of 20 µM in storage buffer and stored at −20 °C.
Nanobody–DNA conjugation
Nanobodies engineered with ectopic amino-terminal and carboxy-terminal cysteines targeting ALFA (clone 1G5, N1502, lot no. 15220402), GFP (clone 1H1, N0302, lot no. 022307) and RFP (clone 2B12, N0402, lot no. 012309) were ordered from NanoTag Biotechnologies and conjugated to azide-modified DNA following established protocols22,58. In brief, nanobodies were diluted in buffer D for 30 min in the dark at 4 °C. After removing TCEP with buffer exchange into PBS using 10 kDa Amicon filters (Sigma-Aldrich, UFC5010; 10 kDa MWCO), A 20-fold molar excess of DBCO-PEG₄-maleimide (Sigma-Aldrich, 760676) was added. Following a 4 h incubation in the dark at 4 °C on a shaker, unreacted crosslinker was removed, and a fivefold molar excess of azide-modified DNA was added for overnight click conjugation at 4 °C. The resulting conjugates (NbALFA-L5R2, NbALFA-LS4, NbALFA-S4, NbGFP-S4 and NbRFP-LS4) were purified from free DNA and unreacted nanobodies using size-exclusion (Superdex 75 10/300 GL GE Healthcare) and anion-exchange (Resource Q 1 ml column Cytiva) chromatography on an ÄKTA Pure system and then stored at −20 °C in storage buffer.
Plasmid construction
The following base plasmids were used: mEmerald-Sec61β, mEmerald-Ensconsin and EGFP-OMP25 (ref. 18). For microtubule labelling, oxStayGold-ALFA-Ensconsin was constructed by replacing the mEmerald sequence in mEmerald-Ensconsin with oxStayGold from pcDNA3/er-(n2)oxStayGold(c4) (Addgene, plasmid no. 185822) and inserting an ALFA tag at the N terminus of Ensconsin. The resulting fragment was cloned into the pSin vector using SpeI and BamHI digestion. For ER or mitochondria labelling, ALFA tags were inserted at the N termini of Sec61β and OMP25 within their respective base plasmids to generate mEmerald-ALFA-Sec61β and EGFP-ALFA-OMP25. Both constructs were subsequently subcloned into the pSin vector for stable cell line generation.
Cell culture and transfection
COS-7 (CRL-1651), HeLa (CCL-2), HEK293T (CRL-3216) and U-2 OS (HTB-96) cell lines were purchased from the American Type Culture Collection. COS-7, HEK293T and HeLa cells were cultured in DMEM (Gibco, C11965500CP) and U-2 OS cells in McCoy’s 5A medium (Gibco, 16600-082). All media were supplemented with 10% FBS (Sigma-Aldrich, F8318) and 1% penicillin–streptomycin (Gibco, 15140122). Before cell seeding, coverslips (Marienfeld, 0117650) and confocal dishes (MatTek, P35G-1.5-14-C) were cleaned by immersion in 1 M KOH (Sigma-Aldrich, 484016) and sonicated for 15 min, subjected to three 10 min sonication cycles in ddH2O, sterilized with 100% ethanol (J&K Scientific, 258449) for 1 min and coated with poly-L-lysine (Sigma-Aldrich, P4707). Approximately 1 × 105 cells were seeded per coverslip for transient transfections with Lipofectamine 3000 (Thermo Fisher Scientific, L3000015) according to the manufacturer’s instructions.
Stable cell line generation
Lentiviruses were produced by co-transfecting HEK293T cells with packaging plasmids psPAX2 (Addgene, plasmid no. 12260) and pMD2.G (Addgene, plasmid no. 12259) along with the appropriate transfer plasmid (pSin-oxStayGold-ALFA-Ensconsin, pSin-mEmerald-ALFA-Sec61β or pSin-EGFP-ALFA-OMP25) using Lipofectamine 3000. Transfection medium was replaced with fresh medium after 12–24 h. At 48–72 h post transfection, viral supernatants were collected, clarified by centrifugation (600g, 5 min), mixed at a 1:4 ratio with a virus concentration reagent (Biodragon, BF06205) and incubated overnight at 4 °C. Viruses were pelleted by centrifugation and resuspended in fresh medium. Target COS-7 and HeLa cells were transduced with the concentrated virus supplemented with 8 µg ml−1 polybrene (Beyotime, C0351). After 4–6 h at 37 °C, the medium was replaced. Following a 24 h recovery, stably transduced cells were selected using 1 µg ml−1 puromycin (Beyotime, ST551) for 3–5 days and expanded for imaging.
Microtubule labelling
COS-7 cells (transfected with oxStayGold-ALFA-Ensconsin for nanobody labelling) were cultured on coverslips. Cells were pre-fixed with pre-warmed 0.3% glutaraldehyde (Electron Microscopy Sciences, 16020) and 0.25% Triton X-100 in cytoskeletal buffer for 2 min, followed by fixation with pre-warmed 2% glutaraldehyde in CB buffer for 10 min. After fixation, cells were quenched with 0.1 M NH4Cl (Sigma-Aldrich, A9434) in PBS for 5 min. After rinsing three times with PBS, cells were blocked and permeabilized with blocking buffer A for 1 h at room temperature. For nanobody labelling, cells were labelled with anti-ALFA nanobodies with docking strands L5R2 or LS4 (NbALFA-L5R2 or NbALFA-LS4) in dilution buffer A at 4 °C overnight. Cells were washed three times with wash buffer for 5 min each, followed by two washes with PBS for 10 min each. Before adding the imager solution (LR2 in buffer C and LP10 in buffer E), samples were washed with buffer C for 5 min. For antibody labelling, cells were incubated with mouse anti-TUBA4A primary antibody (Sigma-Aldrich, clone B-5-1-2, T6074, lot no.0000419295; 1:500) in dilution buffer A, washed three times with wash buffer and subsequently incubated with goat anti-mouse-S4 secondary antibody in dilution buffer A at room temperature for 1 h. After washing, samples were washed with buffer E for 5 min and introduced into the imager solution (P33 in buffer E).
NPC labelling
The U-2 OS NUP96-ALFA-mEGFP CRISPR knock-in cell line was generated in a previous study23. Cells were cultured on coverslips for 48 h, fixed with pre-warmed 2.4% paraformaldehyde (PFA; Electron Microscopy Sciences, 15710) in PBS for 30 min, quenched with 0.1 M NH4Cl in PBS for 5 min and washed four times with PBS for 30 s, 60 s and 2 × 5 min. Permeabilization and blocking were performed simultaneously in blocking buffer B for 2 h at room temperature. For labelling, home-made NbALFA-L5R2 probes (50 nM) in dilution buffer A were added to the samples at 4 °C overnight. After washing, samples were prepared for imaging by adding imager strand LR2 in buffer C.
ER labelling
COS-7 cells stably expressing mEmerald-ALFA-Sec16β were cultured on poly-L-lysine-coated coverslips for 24 h. Cells were fixed with 3% PFA and 0.1% glutaraldehyde in PBS for 15 min at room temperature, quenched with freshly prepared 0.1 M NH4Cl for 5 min. Cells were washed three times with PBS for 5 min each. After incubating with blocking buffer A for 1 h at room temperature, cells were incubated with NbALFA-L5R2 in dilution buffer A at 4 °C overnight. Cells were then washed sequentially with wash buffer (three 5 min washes) and PBS (two 10 min washes), rinsed in buffer C (5 min) and mounted in imager solution (strand LR2 in buffer C).
Two-colour sample labelling
For dual microtubule and ER labelling, COS-7 cells stably expressing oxStayGold-ALFA-Ensconsin were transfected with mEmerald-Sec61β and cultured for 24 h. Samples were processed similarly to ER labelling, except cells were incubated with NbALFA-LS4 and NbGFP-S4 probes in dilution buffer A at 4 °C overnight. After washing, cells were washed with buffer E for 5 min and introduced into the imager solution (LP10 and P81 in buffer E). For dual ER membrane and lumen labelling, COS-7 cells stably expressing mEmerald-ALFA-Sec61β were transfected with mCherry-KDEL11 and cultured for 24 h. Following standard fixation and blocking, cells were incubated with NbALFA-S4 and NbRFP-LS4 probes in dilution buffer A overnight at 4 °C. After washing, cells were washed with buffer E for 5 min and introduced into the imager solution (P10 and LP71 in buffer E).
Mitochondrion labelling
HeLa cells stably expressing EGFP-ALFA-OMP25 and COS-7 cells were processed similarly to samples for ER labelling. For DNA-PAINT, HeLa cells were incubated with NbALFA-LS4 in dilution buffer A overnight at 4 °C. After washing, samples were washed with buffer E for 5 min and introduced into the imager solution (imager strand LP10 in buffer E). For dSTORM, COS-7 cells were incubated with rabbit anti-TOMM20 primary antibody (HUABIO, clone ST04-72, ET1609-25, lot no. H641562031; 1:500) in dilution buffer A, washed three times and incubated with goat anti-rabbit-AF647 secondary antibody (Thermo Fisher Scientific, A21245, lot no. 2390714; 1:1,000) in dilution buffer A at room temperature for 1 h. After washing, samples were imaged in freshly prepared dSTORM buffer containing 1 kU ml−1 catalase (Sigma-Aldrich, C40-100MG), 0.135 kU ml−1 glucose oxidase (Sigma-Aldrich, G2133-50KU) and 35 mM MEA (Sigma-Aldrich, M6500) in base buffer.
Live-cell labelling
U-2 OS cells stably expressing Sec61β-Emerald-HaloTag (a gift from D. Yang, Westlake University) were seeded in confocal dishes. Cells were incubated with 500 µl of PA-JF549-HaloTag ligand (a gift from L. Lavis, Janelia Research Campus) in phenol-red-free DMEM (Gibco, 21063-029) supplemented with 10% FBS for 1 h at 37 °C with 5% CO2. The ligand concentration was 100 nM for live-cell imaging and 25 nM for single-molecule tracking. After three washes with PBS and two washes with live-cell imaging medium (FluoroBrite DMEM; Gibco, A1896701), cells were incubated in the same medium for 1 h at 37 °C to remove unreacted dye. Imaging was performed at room temperature in fresh imaging medium.
Mice
Male C57BL/6J mice (JAX, 000664) were bred in the laboratory or purchased from the animal source centre at Westlake University for synaptonemal complex samples. The Rosa26-GFP-OMP25 knock-in line was generated by inserting a GFP-OMP25 cassette into the Rosa26locus (Shanghai Model Organisms Center). The ChAT-Cre-ChR2-YFP line was generated by crossing ChAT-IRES-Cre (JAX, 006410) and RCL-ChR2(H134R)/EYFP (JAX, 024109). All strains were maintained on a C57BL/6J × ICR background. All experimental procedures were approved by the Institutional Animal Care and Use Committee at Westlake University. Mice were housed in a specific-pathogen-free facility (22–26 °C, 40–70% humidity, 12 h light–dark cycle) with ad libitum access to food and water.
Synaptonemal complex samples
Spermatocyte preparation was adapted from a previous study18. Seminiferous tubules from 3-week-old male C57BL/6J mice were incubated in buffer I (137 mM NaCl, 2.7 mM KCl, 8 mM Na2HPO4, 2 mM KH2PO4; Sigma-Aldrich) for 10 min at room temperature. Released spermatocytes were mixed 1:1 with buffer II (100 mM sucrose, Sigma-Aldrich), passed through a 70 µm strainer and spread onto coverslips presoaked in buffer III (1% PFA, 0.15% Triton X-100, pH 9.2). After drying for 4 h in a humidified chamber and washing, samples were treated with Image-iT (Invitrogen, I36933) for 30 min. Samples were washed three times, blocked with 3% BSA in PBS for 30 min at room temperature and incubated overnight at 4 °C with mouse anti-SCP-3 (Santa Cruz Biotechnology, clone D-1, sc-74569, lot no. J2221; 1:1,000 in 1% BSA/PBS). Following three washes, samples were incubated overnight at 4 °C with goat anti-mouse-S4 (1:500). After a final three washes, samples were mounted in buffer E containing imager strand P10.
Transcardial perfusion and tissue labelling
Solutions were freshly prepared daily. Adult Rosa26-GFP-OMP25 and ChAT-Cre-ChR2-YFP mice were anaesthetized (intraperitoneal 1% sodium pentobarbital, 200 μl) and transcardially perfused at 10 ml min−1 with 50 ml PBS followed by 50 ml 4% PFA in PBS. Brains were excised, post-fixed in 4% PFA at 4 °C for 8–12 h with agitation and rinsed three times for 20 min in cold PBS. Vibratome sections (30 μm; Leica VT1200 S), coronal for Rosa26-GFP-OMP25 mice and sagittal for ChAT-Cre-ChR2-YFP mice, were stored at 4 °C in PBS with 0.05% ProClin 300 (BIOSS, D10200). Before labelling, sections were washed twice in PBS (10 min each), quenched with 100 mM glycine in PBS, washed twice more in PBS and permeabilized or blocked in blocking buffer C for 2 h at room temperature. Sections from Rosa26-GFP-OMP25 mice were incubated overnight at 4 °C with NbGFP-LS4 in dilution buffer B. After three washes in wash buffer and two in PBS (10 min each), sections were washed in buffer E (10 min) and mounted in imager solution (LP10 in buffer E). Sections from ChAT-Cre-ChR2-YFP mice were incubated overnight at 4 °C with rabbit anti-GFP (Thermo Fisher Scientific, A11122, lot 2901498,1:500) in dilution buffer B, washed three times in wash buffer and incubated overnight at 4 °C with donkey anti-rabbit-LS4 (1:500). After identical washing steps, samples were mounted in LP10/buffer E for imaging.
Sample mounting
For fixed-cell imaging, samples were rinsed with fresh imaging solution, placed in a custom holder with 100 μl of imaging solution and covered with a clean coverslip. Excess solution was removed, and the assembly was sealed with two-component silicone glue (Twinsil, Picodent), curing for 10–20 min before imaging. For live-cell imaging, the solution in the confocal dishes was replaced with 300 μl of live-cell imaging medium before covering with a coverslip and sealing in the identical manner.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
