A versatile synaptotagmin-1 nanobody provides perturbation-free live synaptic imaging and low linkage-error in super-resolution microscopy

Imaging of living synapses has relied for over two decades on the overexpression of synaptic proteins fused to fluorescent reporters. This strategy changes the stoichiometry of synaptic components and ultimately affects synapse physiology. To overcome these limitations, here we introduce a nanobody that binds the calcium sensor synaptotagmin-1 (NbSyt1). This nanobody functions in living neurons as an intrabody (iNbSyt1) and is minimally invasive, leaving synaptic transmission almost unaffected, as demonstrated by the crystal structure of the NbSyt1 bound to synaptotagmin-1 and by our physiological data. Its single-domain nature enables the generation of protein-based fluorescent reporters, as we showcase here by measuring spatially-localized presynaptic Ca2+ with an NbSyt1-jGCaMP8 chimera. Moreover, its small size makes the NbSyt1 ideal for various super-resolution imaging methods. Overall, NbSyt1 is a versatile binder that will enable imaging in cellular and molecular neuroscience at a higher precision than possible in the past, over multiple spatiotemporal scales.

Introduction also be functional in living neurons to monitor native Syt-1 location and function with minimal 1 interference on Syt-1 physiological tasks. 2 We present here the successful development and characterization of an alpaca single-3 domain antibody, also known as nanobody (Nb), that remarkably fulfills most of our 4 abovementioned criteria, and therefore it appears as an ideal probe to study living synapses and 5 Syt-1 molecular physiology. This Nb is able to bind with high affinity and specificity to the 6 C2A domain from rat Syt-1 and after obtaining the crystal structure of the Nb bound to rat C2A 7 domain, we predicted that the Nb should significantly minimize the linkage error when directly 8 conjugated to fluorophores and it should also impose a minimal effect on native Syt-1 activity. 9 Due to these already favorable features, we decided to continue engineering this monovalent 10 nanobody as an ideal probe to stoichiometrically reveal Syt-1 in various high-and super-11 resolution microscopy techniques like SIM, STED, DNA-PAINT, and Expansion Microscopy 12 (ExM). In addition, we tested and demonstrated its minimal perturbation when expressing this 13 Nb in living primary neurons (as intrabody), resulting in a flexible tool to label live synapses 14 and follow synaptic vesicles. Finally, we used this specific binder to position the sensitive 15 calcium sensor jGCaMP8 19 directly on pre-synapses, allowing a highly localized calcium 16 detection and fast response to evoked and spontaneous activity without genetic manipulation 17 of synaptic proteins. We conclude that this Nb is a highly versatile tool for use in living 18 neurophysiology studies as in conventional and super-resolution imaging techniques. We 19 expect this Nb to be instrumental in tracking synaptic physiology and dissecting the molecular 20 role of Syt-1 in neuronal communication.

22
Results 23 Selection and initial characterization of potential binders to Syt-1.

24
For the selection of nanobodies against Syt-1 we immunized alpacas with enriched rat 25 synaptosomes. Five months after the immunization protocol, we extracted the peripheral blood 26 mononuclear cells (PBMCs) from 100 ml of blood. After total RNA extraction from the 27 PBMCs we generated a cDNA library of Nbs and cloned them into a phagemid to later screen 28 for binders using phage-display (Supp. Fig. 1a). Before performing the phage-display, we 29 verified if these animals had generated heavy-chain antibodies against the cytosolic (amino 30 acids 97-421) domain of rat Syt-1 (rSyt-1(97-421)). For this, we first removed conventional 31 immunoglobulins (IgG1) that might also recognize this domain and enriched for heavy-chain 32 antibodies (IgG2 & IgG3) from the plasma of the immunized alpacas (Supp. Fig. 1b). These 33 IgG2 and IgG3 were used for the detection of rSyt-1(97-421) coated as antigen in an ELISA plate 34 (Supp. Fig. 1c). The ELISA result suggested that heavy chain antibodies binding rSyt-1(97-421) 35 were present in the serum although we used a complex antigen mixture for immunization 36 (synaptosomes). These results encouraged us to proceed with phage-display screening. 37 After three rounds of phage display screening (biopanning) using the whole 38 immobilized cytosolic rSyt-1(97-421), we obtained five clones with strong positive signals on an 39 ELISA. Although these clones have relatively similar complementary determining region 1 40 (CDR1), they all show significant differences in their CDR2s and CDR3s (Fig.1a, Supp. Fig.  41 1d). After subcloning all five clones into a mammalian expression vector, only two clones (A51 42 and A91) resulted in reasonable protein yields (>20 mg/L). 43 For further validation, NbA51 and NbA91 were directly conjugated with fluorophores 1 for testing them in immunofluorescence (IF) assays. Initially, COS-7 fibroblasts were 2 transfected with plasmids coding for the full-length rat Syt-1 fused to EGFP. After 24h, cells 3 were chemically fixed using paraformaldehyde and IF was performed using 50 nM of 4 fluorescent nanobodies. Both clones displayed specific signals that were detected only on 5 EGFP-positive cells, suggesting that both clones are able to detect rat Syt-1 and that their 6 epitopes seem not to be affected by aldehyde fixation (Fig. 1b). In order to understand the 7 specificity of these candidates, a dot-blot assay with purified rSyt-1(97-421) and rSyt-2(97-422) 8 suggest that both candidates bind specifically to rSyt-1 and not rSyt-2 (Supp. Fig. 2). We 9 carried on an additional validation step to verify their IF performance on rat primary 10 hippocampal cultures. At this point we reasoned that if the nanobody would bind efficiently in 11 a cell context situation, where Syt-1 is in close proximity with several other interacting 12 molecules, it would thus be the one providing the higher signal-to-noise ratio (SNR) likely 13 because has a higher affinity or is able to find more epitopes. While both candidates tested 14 displayed the classical punctuated pattern of synapses expected in these primary neurons, by 15 closer inspection the candidate NbA51 provided images with negligible level of background 16 staining in cell bodies of neurons and astrocytes. Due to this, we decided to continue to work 17 with the NbA51 and rename it NbSyt1.   Nuclei are stained with DAPI and depicted in greyscale.

10
To further confirm the specificity of NbSyt1, we performed the co-staining of primary 11 hippocampal neurons using a directly labeled version of the nanobody and a conventional 12 monoclonal antibody anti-Syt-1, which displayed virtually a complete colocalization on 13 confocal microscopy (Fig. 1d), confirming that this NbSyt1 appears to be also specific in a 14 neuronal environment. Next, we decided to understand more precisely where the NbSyt1 binds 15 within rSyt-1(97-421). For this, we produced various fragments of rSyt-1(97-421) and performed a 16 dot-blot assay. The analysis of these experiments indicated that the NbSyt1 binds to the C2A 17 calcium-binding domain of Syt-1. However, if the putative protein linker between C2A and 18 C2B is removed (fragment-4 & fragment-6, Supp. Fig. 2b) the affinity was strongly reduced or 19 completely lost. This finding motivated us to proceed with the crystallization of the complex 20 NbSyt1-Syt-1 and to resolve their interaction at high resolution. 21 The structure and interaction affinity of the complex NbSyt1 and C2A domain of rSyt- 1 1 In order to perform X-ray crystallography structural studies, the NbSyt1 was expressed 2 and purified in its minimal size and without tags to be mixed in equal molar ratio with the tag-3 free rSyt- for S107 on the NbSyt1 and Q154 and Q209 on C2A(140-265), which make contacts with the 8 backbone of the neighboring chain. As shown in Figure 2d, F40 from the NbSyt1 extends into 9 a hydrophobic pocket on C2A(140-265) and is sandwiched between N207 and Q209. The C2A(140-10 265) aspartate residues 172, 178 and 230 coordinate a calcium ion, and the presence of the 11 NbSyt1 does not seem to affect the calcium binding of the C2A domain. The residue E257 on 12 the crystal structure on the C2A(140-265) has contact with 2 amino acids on the NbSyt1. 13 Interestingly, E257 was missing in fragment-4 when performing the epitope mapping (Supp. 14 Fig. 2b), which could explain the lack of signal of NbSyt1 on fragment-4, but a strong signal 15 with fragment-6 (including E257). Next, we took the full cytosolic domain rSyt-1(97-421) and 16 determined with the microscale thermophoresis assay that the NbSyt1 binds with an affinity of 17 0.7±0.3 nM to rSyt-1(97-421) (Supp. Fig. 3). 18 When analyzing the structure to understand why the NbSyt1 interacts very weakly with 19 rSyt-2, two points were observed. First, the glutamine Q209 in rSyt-1 is exchanged for 20 threonine in rSyt-2. This key interacting residue has direct contact with the backbone of K38 21 and F40 and a polar contact with side chain of residue S107 on the NbSyt1. Secondly, the 22 valine V255 in rSyt-1 is exchanged for proline in rSyt-2, which is expected to alter the 23 backbone geometry of the polypeptide chain and disrupting interactions in its proximity, e.g. 24 the interactions between E257 (conserved in rSyt-2) and the NbSyt1 (S61 & N63). 25 26   2  After determining the NbSyt1 binding affinity, binding epitope and specificity to Syt-3  1, we conjugated the NbSyt1 directly to fluorophores or to single-stranded DNA (ssDNA), to  4  fulfill some of the requirements for various advanced imaging techniques like Stimulated  5 Emission Depletion (STED), Points Accumulation for Imaging in Nanoscale Topography 6 (PAINT) and Structural Illumination microscopy (SIM). Based on the structure, we know that 7 the linkage error from the fluorophore on the N-terminus of the NbSyt1 is only ~1.2 nm to the 8 closest part of the C2A, while the fluorophore located on the C-terminus is slightly below 4 nm 9 from its target. In the case of DNA-PAINT imaging, the Atto488 fluorophore on the imager 10 strand (i.e., the ssDNA annealing to the conjugated ssDNA on the NbSyt1) will be 11 approximately ~4 nm away from Syt-1. This means that for the super-resolution imaging 12 techniques tested here, the small linkage error imposed by the NbSyt1 is neglectable. Results 13 shown on Figure 3 suggest that the commercial STED and SIM setups provided a slightly lower 14 resolution as compared to single-molecule localization-based techniques: Exchange-PAINT 15 20,21 and Fluorescence Lifetime-PAINT (FL-PAINT) 22 strategies. This becomes evident when 16 looking into the gaps in intensity profiles between the Syt-1 localization peak (at the pre-17 synapse) with the postsynaptic density marker PSD-95 peak, also detected with a directly 18 labeled nanobody (Fig. 3bc). Exchange-PAINT had the tendency, in our comparison, to 19 provide a more accurate distance between pre-(Syt-1) and post-(PSD-95) synaptic 20 compartments (i.e., ~100-150 nm 23,24 ). 21

11
Next, we tested if this nanobody could be used in expansion microscopy (ExM). It has 12 been suggested that nanobodies might not work in all ExM techniques 25 . Maybe due to their 13 small size and limited numbers of fluorophores, the signal from nanobodies might not be well 14 retained in some of the hydrogels used (Fig. 4a)

11
Using NbSyt1 as intrabody (iNbSyt1) in living neurons 12 The single-chain nature of nanobodies, allows their expression inside eukaryotic cells.

7
Assessment of synaptic activity in presence of the iNbSyt1 8 To test whether our engineered NbSyt1 intrabody with its high specificity to the C2A 9 domain of Syt-1 is not altering synaptic physiology, we investigated neuronal activity in 10 primary neuronal culture. We first checked changes in synaptic vesicle release properties using 11 the fusion of VAMP2 with the superecliptic GFP, also known as SynaptopHluorin (SpH) 31 . 12 This approach allowed us to evaluate if the iNbSyt1 has any effect on synaptic vesicle release 13 and SV endocytosis. First, we confirmed that the iNbSyt1 is clearly localized at presynaptic 14 terminals where it strongly colocalized with immunostainings for Synaptophysin and Bassoon 15 (Fig. 6a). Then, we generated a bicistronic AAV expressing the iNbSyt1 and SpH 31 . When 16 following the SpH signal after 60 action potentials (AP) or 600 AP at 20 Hz stimulation rate, 17 we were not able to observe differences in SyH signal traces between neurons with or without 18 expression of the iNbSyt1 (Fig. 6b). These results directly suggest that the expected enrichment

4
Next, we looked for changes in the spontaneous release rate that could potentially arise 5 from expression of our Syt-1 intrabodies. We therefore recorded non-evoked excitatory and 6 inhibitory postsynaptic currents in the presence of tetrodotoxins in primary neuronal culture 7 (miniature EPSCs and IPSCs) (Fig. 7a) for both, the iNbSyt1 and the original candidate 8 iNbA91. We observed no significant differences in the amplitudes (quantal release of 9 neurotransmitters) nor in Tau(on) and Tau(off) kinetics between cultures infected with iNbSyt1 10 and iNbALFA 32 (used as a control intrabody since it has no target molecules in neurons). For 11 the mEPSC rate, we observed a trend for both iNbSyt1 (iNbA51) and iNbA91 with a slight 12 increase in rate of miniature postsynaptic currents, however not statistically significant for 13 iNbSyt1 (Fig. 7d). 14 15

19
Boxplots summarizing mIPSCs and mEPSCs properties such as rate and amplitude (b and d) and Tau(on) 20 and Tau(off) (c and e). For statistics, mIPSCs, n = 10-12 cells per group; mEPSCs, n = 18-21 cells per 21 group, a one-way ANOVA with multiple comparison was applied with Tukey correction. P-values:

24
In light of our structural data that shows binding of NbSyt1 to the C2A domain of Syt-25 1, these findings are in line with the observations made by Courtney et al. 8 . They also observed 26 that after genetically removing the C2A domain from Syt-1, a minimal effect on evoked release 27 and a slight increase in the spontaneous miniature release could be observed. On the contrary, 28 they also show that if the C2B domain is truncated, the action potential triggered synaptic 29 vesicle fusion to the plasma membrane is drastically impaired 8 . Therefore, our data suggest 30 that the iNbSyt1 bound to the C2A has no detrimental physiological effect in evoked activity 31 on the synaptic physiology, and potentially neglectable effects in facilitating spontaneous 32 release.

34
Targeted calcium detection on pre-synapses 1 After characterizing the nanobody, its interactions and its effects on synaptic 2 physiology, we thought to use it as an ideal tool to increase the synaptic targeting of genetically 3 encoded calcium indicator, to be able to measure the Ca 2+ changes in close proximity of the 4 SV sensor for Ca 2+ (i.e., Syt-1). This approach would allow us to reveal active synapses in 5 living neurons, and perform calcium imaging on pre-synapses with high precision and with 6 minimal impact on neuronal physiology, as we are not overexpressing chimeric synaptic 7 vesicle proteins for the targeting. For this we generated new AAVs expressing the NbSyt1 as 8 an intrabody and fused to the calcium sensor jGCaMP8s and jGCaMP8m (iNbSyt1-9 jGCaMP8) 19 . As control, we used the iNbALFA fused to the calcium sensors. As the NbALFA 10 has been shown to work as intrabody 32 , but has no specific binding partners in rat hippocampal 11 neurons, this would be the most appropriate control for these experiments. As expected from 12 previous results, we observed the enrichment of synaptically-targeted calcium sensors enriched 13 at synaptic boutons (Supp. Movie 3), and mostly a diffused signal with the control (Fig. 8a). 14 Both sensors showed good signal following evoked 1 and 3 APs (Fig. 8b). However, when 15 observing the traces from single boutons, the synaptically localized sensor provided more 16 reliable signals, as single action potentials were sometimes lost in the background when 17 measuring with the control sensor (Fig. 8b). To further corroborate this observation, we 18 performed a more formal analysis on several boutons, also considering if the synaptically 19 targeted sensor provides additional spatial information for the calcium signal compared to the 20 control. For this we identified the boutons at the end of each movie following a 300 AP 21 stimulus, and used this signal to select the center of the boutons synaptic proximal and distal 22 areas (Fig. 8c). Using this analysis, we were able to see that the pre-synaptically localized 23 sensors provided more precise spatial resolution and better signal to noise ratio (SNR) when 24 compared to the control (Respectively Fig. 8d, e and Fig. 8f, g). Finally, having this highly 25 sensitive and localized calcium sensor, allowed us to follow spontaneous network activity and 26 observe bursts on primary hippocampal neurons using the slow and medium speed sensors 27 (Fig. 8h, i).

6
Discussion 7 Here we developed and characterized an extremely flexible affinity tool, the NbSyt1, 8 that allows the study of Syt-1 from several angles. We prove the usefulness of this Nb in 9 conventional microscopy, various super-resolution imaging modalities, and as an intrabody in 10 living neurons to follow the molecular physiology of synapses without overexpressing a tagged 11 synaptic protein. 12 We selected a camelid single-domain antibody binding with excellent specificity and 13 affinity to the C2A calcium-binding domain of Syt-1. The structural data we obtained from the 14 Syt-1-NbSyt1 complex confirms the very high binding affinity (~0.7 nM) and specificity, 15 which is a consequence of polar and hydrophobic interactions, explaining its selectivity for 16 Syt-1. NbSyt1 does not recognize the C2A domain of Syt-2, which shares a very high degree 17 of identity with Syt-1 (77,7% identity on the full protein, and 86,6% on the C2A domain). 18 Importantly, the crystal structure also shows that the binding occurs on the opposite side from 19 the calcium-binding pocket of the C2A domain, which encouraged us to use this nanobody in 20 living neurons as an intrabody, as this region is likely to be not involved in lipid or other protein 21 interactions 10,14 . 22 The advantages of nanobodies as affinity probes for imaging over classical antibodies 23 have been thoroughly described 24,33,34 . Some of the main benefits are their monovalency 24 (stoichiometric labeling) and their small size allowing to place the fluorophore ~2-4 nm from 25 the target. These features make NbSyt1 ideal for quantitative imaging 35-37 and high-precision 26 nanoscopy, respectively, where conventional antibodies would place the fluorophores with a 27 linkage error comparable to or sometimes larger than the resolution that different techniques 28 attain 24,38 . 29 Additionally, the NbSyt1 is functional when expressed in neurons as an intrabody 30 (iNbSyt1). This can be clearly observed in vivo as the iNbSyt1 is able to get enriched on 31 presynaptic boutons with practically no effect on SV release physiology (Fig. 6). This feature 32 results in the possibility of specifically labelling synapses relying on the endogenous 33 expression levels of Syt-1 without changing the stoichiometry of synaptic proteins by over-34 expression or modifying their function or location by fusing them with tags like fluorescent 35 proteins. Therefore, we demonstrated using SpH (Fig. 6) and electrophysiology (Fig.7 matter the most for SV-release regulation, in close proximity to its sensor, results in the precise 45 and locally-enhanced detection of a single AP with an increased signal-to-noise ratio. 1 Similarly, we used this tool for recording the spontaneous activity of hippocampal neuronal 2 cultures using two chimeras with different speeds in detecting Ca 2+ variations (respectively the 3 slower iNbSyt1-jCaMP8s and the faster iNbSyt1-jCaMP8m). These two tools will serve the 4 community depending on different experimental needs both for the study of presynaptic 5 calcium and for the role of Syt-1 in different cellular models. 6 Overall, here we established, engineered and thoroughly characterized a nanobody 7 binding to the C2A domain of Syt-1. This versatile tool shows outstanding features for 8 quantitative and super-resolution microscopy, but it also provides exceptional properties to be 9 used in living neurons and follow or manipulate presynaptic molecular physiology. We expect 10 this bio-tool to be instrumental in understanding Syt-1's role in SV fusion and neuronal 11 physiology. 12 13 Material and Methods 1 2 Synaptosomes preparations. Rat synaptosomes were enriched as previously described 39 3 Briefly, rat brains were homogenized using a glass-Teflon homogenizer in precooled sucrose 4 buffer (320 mM Sucrose, 5mM HEPES, pH 7.4). Centrifugation at 1000 x g for 2 minutes was 5 performed, and the supernatant was further centrifuged at 15,000 x g for 12 minutes. Next, a 6 discontinuous Ficoll density gradient was applied. The fractions at the interface of the 9% 7 Ficoll were pooled and washed in sucrose buffer. 8 9 Immunizations. Two alpacas were immunized with this preparation. The procedure was 10 performed by Preclinics GmbH (Postdam, Germany). Six injections were performed weakly 11 with 500 µg rat synaptosomes (total protein determined by BCA assay). Two weeks after the 12 last immunization a single boost with 500 µg of synaptosomes was performed and 100 ml of 13 blood was taken 3 and 5 days after the boost immunization. PBMCs were isolated using Ficoll 14 gradient and Serum was stored at -80°C. Total RNA was extracted using RNA extraction 15 Qiagen kit (Qiagen Nanobody library generation. Total mRNA was extracted from the PBMC obtained from 2 4 alpacas using standard RNA extraction kit (Qiagen). Recovered mRNA was retrotranscribed 5 to cDNA by using Supercript IV (ThermoFischer) and the Cal 0001/2 primers as described 6 before 41 . Next, a second PCR was performed to introduce the Gibson cloning overhangs for 7 further insertion in the phagemid. The final PCR product was diluted to 5 ng/µL and some 8 loaded on a 1.5 % Agarose gel to confirming the right size of the PCR product. Fragments were 9 cloned into the phagemid using Gibson assembly. The phagemid backbone used was obtained 10 from the pHen2 plasmid. After Gibson cloning, the obtained construct was purified by PCR 11 purification kit (Qiagen) and the concentration was measured by Nanodrop. The constructs 12 were then electroporated in TG1 bacteria (Biocat). For the transformation, 65 ng of DNA were 13 added to 50 µL of TG1. This process was repeated 20 times. The reactions were left 1 hour at 14 37°C for initial growth and then pooled in 400 mL of 2YT medium (ThermoFischer) 15 supplemented with antibiotics and grown overnight at 37°C. Next day, bacteria were pelleted 16 and resuspended in 25 mL LB medium (ThermoFischer) and 25 % Glycerol. The 17 Synaptosome-library was aliquoted, snap frozen and stored at -80 °C. 18 19 Phage-Display. The procedure was performed as described in Maidorn et al. 42 with some 20 modifications. To start the process, a 1 mL the synaptosome nanobody library was diluted in 21 500 mL of 2YT supplemented with antibiotic and grow at 37°C until OD600 reached ~0.5. 22 Next, ~0.1x10 13 M13KO7 Helper Phages (NEB) were added to the culture and let the infection 23 for 45 minutes at 37°C. Bacteria were then pelleted and resuspended in 500 mL 2YT medium 24 supplemented with the necessary antibiotics. Infected bacteria were incubated overnight at 25 30°C to produce the phages. The next day culture supernatant was incubated with 4 % (w/v) 26 PEG-8000 and let on ice for <2 hours to allow phages to precipitate. After several washings 27 steps in PBS, phages were filtered with a 0.45 µm syringe filter (Sartorius). Purified Syt-1(97-28 421) was conjugated to desthiobiotin-N-Hydrosuccimide Ester (Beryy and Associates). Excess 29 of dt-Biotin-NHS was removed using Nap10 column (Cytiva with PBS-T and blocked using 5% skim Milk and 1% BSA in PBS-T for 2h at RT. Next, 25 µl 2 of phages produced by the picked colonies were incubated on the well with 75 µL 5% skim 3 milk in PBS-T for 1 hour at RT. The unbound phages were washed six times with PBS-T, and 4 bound phages were detected with anti-major coat protein M13-HRP (Santa Cruz, #sc-53004-5 HRP) diluted 1:1000 in 100 µL PBS-T. The antibody was let for 1 hour at RT and excess was 6 thoroughly washed away with PBS-T. TMB substrate (3,3',5,5'-tetramethylbenzidine, 1-7 StepTM Ultra TBM-ELISA, Thermo Scientific) was added to each well and the colorimetric 8 reaction was stopped by addition of 100 µL of 2M sulfuric acid. The absorbance was read at 9 450 nm throughout the plate (Cytation TM 3: BioTek TM Instruments, Inc.). 10 11 Protein expression and purification. Syt-1 fragments were produced in NEB-Express (NEB), 12 while nanobodies were produced in SHuffle® Express (NEB Fluorophore conjugation to nanobodies. Purified nanobodies bearing one or two ectopic 29 cysteines (at their C-terminus or N-and C-termini) were reduced for 1 h using 10 mM of tris  carboxyethyl)phosphine (TCEP), pH 7. The excess of TCEP was removed using a gravity 31 NAP5 column (Cytiva) previously equilibrated with degasses PBS pH 7.4. Freshly reduced 32 nanobodies were immediately mixed with ~3-5 molar excess of maleimide functionalized 33 fluorophore and incubated for 1h. The excess of dye was removed by using Superdex™ 75 34 increase 10/300 GL column (Cytiva) on Äkta FPLC system. 35 36 Oligonucleotide conjugation to nanobodies. NbSyt1 was produced having one ectopic 37 cysteine on its C-terminus, while secondary anti-mouse IgG1 nanobody and anti PSD-95 38 nanobody carrying an ectopic cysteine were obtained from NanoTag Biotechnologies GmbH 39 (#N2005 and #N3705). Nanobodies used for DNA-PAINT imaging were coupled with 40 different docking single-DNA-strands (Biomers GmbH, Ulm, Germany) as described 41 earlier 21,22 . In brief, nanobodies were reduced with 5 mM TCEP, pH: 7, (Sigma-Aldrich, 42 #C4706) for 2 h. After removal of TCEP via 10 kDa molecular weight cut-off (MWCO) 43 Amicon spin filters (Merck, #UFC500324), the reduced nanobodies were coupled using 10-44 fold excess of maleimide-DBCO crosslinker (Sigma-Aldrich, #760668). The excess of 1 crosslinker was removed using 10 kDa MWCO Amicon spin filters. Azide functionalized 2 DNA-strands (Biomers) were mixed in excess with nanobody-DBCO to promote the azide-3 alkyne cycloaddition (click) reaction. The excess Azide docking DNA-strands were removed 4 by size exclusion chromatography using an Äkta pure 25 system (Cytiva) equipped with the 5 Superdex® Increase 75 column (Cytiva). The sequences of the docking strands P3 (3'-6 TTTCTTCATTATTTT-5') and R4 (3'-ACACACACACACACACACA-5') 43,44 7 8 Dot blot assays 9 Target proteins and negative control (BSA) were spotted (~1 µg and ~5 µg respectively) on a 10 nitrocellulose membrane and let them dry at RT. Membranes were incubated with blocking 11 buffer (PBS supplemented with 0.05% (v/v) of Tween20 and 5% milk) under gentle shaking 12 for 2h. After blocking step, blocking buffer was removed and membranes were incubated with 13 NbA51 or NbA91 directly labeled with AbberiorStar635p in 5% milk PBS-Tween20 at a final 14 concentration of 5 nM for 1h. For specificity assay in Supp. Rat primary hippocampal neuron cultures for imaging were prepared as described before 28 . In 27 brief, the brains of P1-2 rat pups were extracted and placed in cold HBSS (ThermoFisher). The 28 hippocampi were extracted and placed in a solution containing 10 mL DMEM (ThermoFisher), 29 1.6 mM cysteine, 1 mM CaCl2, 0.5 mM EDTA, 25 units of papain per mL of solution, with 30 CO2 bubbling, at 37°C for 1h. The solution was removed and the hippocampi were incubated 31 in 10% FBS-DMEM, 73 µM albumin for 15 minutes. The hippocampi were triturated using a 32 10 mL pipette in complete-neurobasal medium (Neurobasal A (ThermoFisher), containing 2% 33 B27 (ThermoFisher) and 1% Glutamax-I (ThermoFisher)). Neurons were plated (12-well 34 plate) on glass coverslips coated with poly-L-lysin-hydrochloride (1 mg/ml, Merck). After 2h, 35 the plating medium was replaced with 1.25 ml complete-neurobasal medium and neurons were 36 incubated for 15 days at 37°C, 5% CO2 in a humidified incubator. Cultures for 37 electrophysiology recordings were prepares according to Goslin and Banker 45 using embryonic 38 day 18 old rat embryos. Cells were plated in a density of 40.000 cells per 18 mm coverslip, 39 grown in 1 ml of neurobasal medium (NB, Gibco) supplemented with B27 medium. Ara-C 40 (cytosine β-D-arabinofuranoside) at a final concentration of 5 µM was included in the culture 41 medium at DIV 7 to suppress glia proliferation. 42 43 Affinity determination. The affinity of the NbSyt1 was measured by microscale 1 thermophoresis using the device NT.115Pico Monolith (NanoTemper). NbSyt1 was labelled 2 with Alexa647 (as described above) and diluted in MST buffer (NanoTemper) supplemented 3 with 0.05 % Tween. Fluorescent NbSyt1 was incubated with different dilutions of purified Syt-4 1(97-421) using Premium Coated Capillaries (NanoTemper). For operation of the instrument and 5 evaluation of affinity data, the MO.Control and MO.Affinity Analysis software (NanoTemper) 6 were used. 7 8 Immunofluorescence. For Fig. 1b COS-7 Cells were fixed for 20 minutes using 4% 9 paraformaldehyde (PFA) in PBS at RT. After rinsing short with PBS, remaining aldehyde 10 groups were quenched for 15 minutes using 0.1 M glycine in PBS at RT. Cells were blocked 11 and permeabilized with 3% (w/v) BSA + 0.1% (v/v) Triton X-100 for 40 minutes at RT and 12 gently shaking. Nanobody candidates were applied in PBS supplemented with 1.5% BSA and 13 0.05% Triton X-100 for 2h at RT with gentle shaking. After staining, several washing steps 14 using PBS were carried out including DAPI short staining. Coverslips were shortly rinsed in 15 distilled water and mounted using Mowiöl (12 ml of 0.2 M Tris buffer, 6 ml distilled water, 6 16 g glycerol, 2.4 g Mowiol 4-88, Merck Millipore). Samples were imaged directly or within the 17 next 48h, samples were kept at 4°C. Primary hippocampal neuronal cultures (15 DIV) imaged 18 in Fig. 1c and Fig.3  JIR; for Fig. 1) or for Fig. 3a and g, the anti-α-Tubulin primary Ab ((#302211, SySy) was pre-24 mixed with FluoTag-X2 anti mouse IgG1 coupled to Atto488 (#N2002, NanoTag 25 Biotechnologies) initially in 20 μl of PBS and after 30 minutes at RT, 980µl of PBS was added 26 to bring the concentration of Ab and nanobody to 7 and 15 nM respectively. Into this pre-27 mixture tube, 25 nM of FluoTag-X2 anti-PSD95 coupled to AbberiorStar580 (#N3702, 28 NanoTag Biotechnologies) and 25 nM NbSyt1 coupled to AbberiorStar635p were added. 29 Samples were incubated for 1 hour at RT under slow orbital stirring. Neurons were washed 3x 30 with PBS and rinsed once with high salt PBS (0.5 M NaCl). Samples were finally rinsed in 31 water and mounted on glass slide using Mowiol. For DNA-PAINT imaging (Fig. 3b, c, d, e  32 and f) immunostaining was performed using ~50 nM of the NbSyt1 functionalized with the 33 docking DNA strand P3 coupled to Atto488 on its 3'-end (NbSyt1-P3-Atto488). 34 Simultaneously, post-synapses were labelled with 50 nM of FluoTag-X2 anti-PSD95 (#N3702, 35 NanoTag Biotechnologies) functionalized with the docking DNA strand R4 on its 3'-end 36 (NbPSD95-R4). After immunostaining, samples were post-fixed with 4% PFA for 15 minutes, were processed using XDS 47 , molecular replacement was performed with Phaser 48 using 3 structures 5T0R 49 and 6I2G 32 as search models. The structure was built using Coot 50 and 4 refined in Phenix 51 . Data collection and refinement statistics are shown in Supp. Table 1.  5  6 Live intrabody imaging and electrical stimulation in hippocampal neurons. Images were 7 taken by an inverted Nikon Ti epifluorescence microscope (Nikon Corporation, Japan) 8 equipped with a Plan Apochromat 60×, 1.4 NA oil immersion objective, an IXON X3897 9 Andor camera, an Oko Touch Environmental control and NIS-Elements software. of spontaneous network activity, neurons were imaged for 60s in the absence of blockers. For 30 responses following field stimulation, the Tyrode's solution was supplemented with 10 μM 31 CNQX and 50 μM D-AP5 to avoid spontaneous network activation. Neurons underwent field 32 stimulation at 20Hz using one field stimulation (one action potential, 1AP) or 3APs. Images 33 were taken at a rate of 15 fps. 34 The analysis of images with these sensors was performed using Fiji/imageJ 52  fused to a C-terminal ALFA-tag and cloned into an AAV backbone 54 , which additionally 40 contained an expression cassette for Synpatobrevin2-pHluorin (SynaptopHluorin or SpH). 41 Both genes, the intrabody and the SpH are under the control of a hSyn1-promoter. For the 42 constructs used in calcium imaging, the first hSyn1-promoter together with the Vamp2-43 pHluorin cassette was removed. Subsequently, the desired sequence for jGCaMP8s/m was 44 cloned in frame to NbSyt1-ALFA or NbALFA-3xFLAG by restriction sites for AgeI and AscI. 1 For live imaging of the intrabodies NbSyt1 (iNbSyt1) and NbALFA (iNbALFA), the nanobody 2 sequences were fused to mScarlet or mNeonGreen. Fluorescent proteins were amplified with 3 the following primers (forward: TCCACCGGTACCATGGTGAGCAAGGGCGAGG and  4 reverse: ATTGGCGCGCCTCTTGTACAGCTCGTCCATGCC) to introduce the restriction 5 sites AgeI and AscI and cloned into the AAV backbone containing the iNbSyt1-ALFA or 6 iNbALFA-3xFLAG. 7 8 Adeno-associated viruses (AAVs). The AAVs were generated as previously described 54,55 . 9 AAVs were produced in HEK293T cells by co-transfection of pHelper plasmids (pFΔ6, pRV1, 10 p21) and pAAV target plasmid in a 4:1:1:2 molar ratio by use of Lipofectamine 2000 11 (ThermoFisher) according to company protocol. 72 h post transfection, cells were harvested 12 and lysed by 3 cycles of thawing and freezing followed by treatment with benzoase nuclease 13 (Millipore MgATP, 0.3 mM of GTP, 0.6 mM of EGTA, and 10 mM of QX314. Whole-cell recordings 32 were carried out at -60 mV (reversal potential of GABAA receptors) for mEPSCs and 0 mV 33 (reversal potential of AMPA and NMDA receptors) for mIPSCs. For statistical analysis (Fig.  34  7), two-tailed unpaired Mann-Whitney U test was performed. It is a nonparametric test in which 35 the sample medians are compared based on the shape of distribution between the two 36 independent groups. To correct significance level, a Bonferroni correction was then employed. 37 The resulting box plots display the median and the first and third quartile with whiskers 38 extending to 1.5 times the interquartile range above and below the upper and lower quartiles, 39 respectively (Q1 -1.5 * IQR or Q3 + 1.5 * IQR