sign in sign up
<<
Home , AAK1, CAMK2B, CAMK2G, CDC42BPA, DCLK1, MINK1

Hidden proteome of synaptic vesicles in the mammalian brain

Zacharie Taoufiqa,1, Momchil Ninovb, Alejandro Villar-Brionesc, Han-Ying Wanga, Toshio Sasakid, Michael C. Royc, Francois Beauchaina, Yasunori Morie, Tomofumi Yoshidae, Shigeo Takamorie, Reinhard Jahnb,1, and Tomoyuki Takahashia,1

aCellular and Molecular Synaptic Function Unit, Okinawa Institute of Science and Technology Graduate University, Okinawa 904-0495, Japan; bDepartment of Neurobiology, Max Planck Institute for Biophysical Chemistry, D-37077 Göttingen, Germany; cInstrumental Analysis Section, Okinawa Institute of Science and Technology Graduate University, Okinawa 904-0495, Japan; dImaging Section, Okinawa Institute of Science and Technology Graduate University, Okinawa 904-0495, Japan; and eLaboratory of Neural Membrane Biology, Graduate School of Brain Science, Doshisha University, 610-0394 Kyoto, Japan

Contributed by Reinhard Jahn, November 11, 2020 (sent for review June 15, 2020; reviewed by Yukiko Goda, Stefan Hallermann, and August B. Smit) Current proteomic studies clarified canonical synaptic that proteomics, combined with subcellular fractionation, yields pro- are common to many types of . However, proteins of tein inventories of high complexity. For instance, >2,000 diversified functions in a subset of synapses are largely hidden species were identified in (1), ∼400 in the SV fraction because of their low abundance or structural similarities to abun- (2), ∼1,500 in postsynaptic densities (3), and ∼100inanAZ- dant proteins. To overcome this limitation, we have developed an enriched preparation (4). “ultra-definition” (UD) subcellular proteomic workflow. Using pu- While these studies provide insights into the protein compo- rified (SV) fraction from rat brain, we identified sition of synaptic structures, they are still inherently limited for 1,466 proteins, three times more than reported previously. This two reasons. First, synapses are functionally diverse with respect refined proteome includes all canonical SV proteins, as well as nu- to the chemical nature of their , as well as their merous proteins of low abundance, many of which were hitherto synaptic strength, kinetics, and plasticity properties (5). There- undetected. Comparison of UD quantifications between SV and syn- fore, analyzed subcellular fractions represent “averages” of a aptosomal fractions has enabled us to distinguish SV-resident pro- great diversity of synapses (6) or SVs (2). The second limitation teins from potential SV-visitor proteins. We found 134 SV residents, is that proteins known to be present in specific subsets were not of which 86 are present in an average copy number per SV of found in these studies, despite the unprecedented sensitivity of less than one, including vesicular transporters of nonubiquitous modern mass spectrometers. In fact, many functionally critical neurotransmitters in the brain. We provide a fully annotated re- synaptic proteins have remained undetected. For example, the source of all categorized SV-resident and potential SV-visitor pro- (Syt) family, major Ca2+ sensors of SV , teins, which can be utilized to drive novel functional studies, as we > characterized here Aak1 as a regulator of synaptic transmission. comprises 15 members, of which only 5 had been identified in Moreover, proteins in the SV fraction are associated with more than previous SV proteomics (2, 4, 7). Missing isoforms included Syt7, 200 distinct brain diseases. Remarkably, a majority of these proteins involved in asynchronous transmitter release (8), synaptic plas- was found in the low-abundance proteome range, highlighting its ticity (9), and SV recycling (10). Likewise, the vesicular trans- pathological significance. Our deep SV proteome will provide a fun- porters for monoamines (VMATs) and (VAChT) damental resource for a variety of future investigations on the func- tion of synapses in health and disease. Significance

| deep proteomics | synaptic vesicles | brain disorders | Mammalian central synapses of diverse functions contribute to highly complex brain organization, but the molecular basis of syn- aptic diversity remains open. This is because current synapse pro- he functions of eukaryotic cells, in all their complexity, depend teomics are restricted to the “average” composition of abundant Tupon highly specific compartmentalization into subcellular synaptic proteins. Here, we demonstrate a subcellular proteomic domains, including organelles. These compartments represent workflow that can identify and quantify the deep proteome of functional units characterized by specific supramolecular protein synaptic vesicles, including previously missing proteins present in a complexes. A major goal of modern biology is to establish an ex- small percentage of central synapses. This synaptic vesicle proteome haustive, quantitative inventory of the protein components of each revealed many proteins of physiological and pathological relevance, intracellular compartment. Such inventories are points of depar- particularly in the low-abundance range, thus providing a resource ture, not only for functional understanding and reconstruction of for future investigations on diversified synaptic functions and biological systems, but also for a multitude of investigations, such neuronal dysfunctions. as evolutionary diversification and derivation of general principles Author contributions: Z.T. designed research; Z.T., M.N., A.V.-B., H.-Y.W., T.S., M.C.R., F.B., of biological regulation and homeostasis. Y.M., and T.Y. performed research; M.N. contributed new reagents/analytic tools; Z.T. and Essential to communication within the , chem- A.V.-B. analyzed data; Z.T., S.T., R.J., and T.T. wrote the paper; R.J. provided technical ical synapses constitute highly specific compartments that are support; and R.J. and T.T. provided supervision. connected by to frequently distant neuronal bodies. Reviewers: Y.G., RIKEN Center for Brain Science; S.H., Leipzig University; and A.B.S., Fac- Common to all chemical synapses are protein machineries that ulty of Science, Vrij Universiteit. orchestrate exocytosis of synaptic vesicles (SVs) filled with neu- Competing interest statement: R.J. (author) and A.B.S. (reviewer) are coauthors on a 2019 SynGO consortium article [F. Koopmans et al., 103, 217–234 (2019)]. rotransmitters in response to presynaptic action potentials (APs), This open access article is distributed under Creative Commons Attribution License 4.0 resulting in activation of postsynaptic receptors. Moreover, syn- (CC BY). apses are composed of structurally and functionally distinct sub- 1To whom correspondence may be addressed. Email: [email protected], rjahn@ compartments, such as free and docked SVs, endosomes, active gwdg.de, or [email protected]. zones (AZs) at the presynaptic side, and receptor-containing This article contains supporting information online at https://www.pnas.org/lookup/suppl/ membranes with associated scaffold proteins on the postsynaptic doi:10.1073/pnas.2011870117/-/DCSupplemental. side. Thus, it is not surprising that mass spectrometry (MS)-based First published December 21, 2020.

33586–33596 | PNAS | December 29, 2020 | vol. 117 | no. 52 www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Downloaded by guest on October 1, 2021 neurotransmitters were missing in these studies. Clearly, known the resolving power of the UD method is based upon improved components of the diversified synaptic proteome have been missing, workflow prior to MS analysis (SI Appendix, Fig. S1). Note that and it is not possible to predict how many more such proteins each sample used for our MS analyses was checked by electron remain hidden. microscopy (EM) and electrophoresis. Typical synaptosomal What are the reasons for the continuing incompleteness of the profiles were observed in P2′ samples whereas uniform vesicle synaptic protein inventory? Proteome identification and quanti- structures of 40 to 50 nm in diameter predominated SV fractions fication rely heavily on MS detectability of peptides generated by (Fig. 1E). Proteins extracted from the P2′ and SV fractions digestion of extracted proteins with sequence-specific , showed distinct sodium dodecyl sulfate polyacrylamide gel such as trypsin. However, in MS analysis of complex biological electrophoresis (SDS/PAGE) profiles (Fig. 1F). samples, peptide signals from a few abundant proteins often mask those that are less abundant. Additionally, the probability of Improved Quantification Revealed the Synaptic Organization and obtaining peptides with similar masses, but different Diversity of the SV Proteome. In quantitative MS, protein abun- sequences, increases with increasing sample complexity (11, 12). To dance can be determined using intensity-based absolute quanti- overcome these limitations, we have elaborated a workflow with fication (iBAQ), a label-free approach in which the summed dual-enzymatic protein digestion in sequence combined with an intensities of all unique peptides of a protein are divided by the extensive peptide separation prior to MS analysis. As proof of total number of unique peptides detected. Thus, the increased concept, we have utilized purified SV fractions from rat whole peptide recovery achieved with the UD method is expected to brain, which serve as a benchmark for quantitative organellar improve the accuracy of protein quantification. To test this as- proteomics (2). As a result, we detected ∼1,500 proteins in the SV sumption, we performed immunoblot analyses for 41 proteins in fraction, three times more than reported previously. This proteome the fractions during SV purification and compared with the not only covers all known canonical SV proteins but also contains quantification profiles of the HD and UD methods (SI Appendix, proteins previously overlooked, such as the low-abundance Syts Fig. S2). As expected, proteins located at the postsynaptic side or and SV transporters. Moreover, peptide quantification allowed for in the synaptic cleft were found in the P2′ fraction, but not in the “ ” “ ” differentiating SV-resident from SV-visitor proteins. In fact, SV fraction, both in immunoblot and MS analyses (SI Appendix, “ ” most SV-resident proteins revealed in our SV proteomics are of Fig. S2A). Proteins residing on SVs were found at higher levels in low abundance, with an average copy number of less than 1 per SV, the SV than the P2′ fraction, in both Western blot and UD suggesting a larger molecular and functional diversity of SVs than analyses (SI Appendix, Fig. S2B). In contrast, the HD method previously thought. Remarkably, more than 200 proteins detected failed to detect some SV proteins in P2′. Similar inconsistencies in the SV fraction are genetically associated with brain disorders, between HD iBAQ data and immunoblot profiles were found for NEUROSCIENCE 76% of which were previously hidden. proteins in AZ, presynaptic membrane, and cytoplasm (SI Ap- pendix C–E Results , Fig. S2 ): altogether in 30% (13 of 41) of cases. These results highlight the importance of the UD workflow for A Workflow with Enhanced Peptide Recovery and Separation Greatly quantitative proteomics. Extended Synaptic Proteome Coverage. A workflow was developed ′ “ SVs are purified from synaptosomes (P2 ), which contain all SV to increase coverage of protein-specific sequences or unique proteins, whereas SVs may not contain proteins from other syn- peptides” prior to MS identification. First, to increase the aptic compartments. Of 4,424 proteins in the P2′ fraction, 3,005 number of accessible cleavage sites, we introduced Lys-C treat- were detected only in P2′, including postsynaptic and mitochon- ments before and during tryptic digestion (Fig. 1A and SI Ap- drial proteins (Fig. 2A). Of 1,466 SV proteins, 1,419 were detected pendix, Fig. S1). Second, to improve separation of the peptides, in P2′. The remaining 47 SV proteins were of low abundance, we introduced off-line fractionation using electrostatic repulsion- including VGLUT3, a vesicular isoform hydrophilic interaction chromatography (ERLIC), based on their charges, polarities, isoelectric pH, posttranslational modifica- present in a limited set of (CNS) synapses. tions, and orientations (13, 14), prior to conventional To evaluate possible contamination of postsynaptic proteins into hydrophobicity-based reverse-phase chromatography (RPC). To the SV fraction, we have referred to the Synaptic Ontologies evaluate the contribution of this workflow to greater protein cover- (SynGO) resource (15). Of all SV fraction proteins, 97 (7%) are age, we also ran a conventional protein digestion-peptide separation annotated as postsynaptic proteins, but 47 out of 97 proteins are protocol combined with modern mass spectrometer (Q-Exactive reportedly present and function in presynaptic compartments Plus) analyses, which we designated as the “high-definition” (HD) (Dataset S1). Thus, contamination of postsynaptic proteins in the method, whereas we refer to our workflow as the “ultra-definition” SV proteome seems minor within the proteins detected in the (UD) method (Fig. 1B and SI Appendix,Fig.S1). The UD-based SynGO database. proteomics revealed 1,466 proteins in the SV fraction (Fig. 1C). This To distinguish SV residents from proteins transiently inter- “ ” is twice as many as with the HD method (766), and more than three acting ( visitors ) with SVs, we determined the iBAQ ratio SV/ ′ B times as many as previously reported (2). The increased sensitivity of P2 in a volcano plot (Fig. 2 ). Of 1,466 SV proteins, 134 had an ′ P < the UD method is also evident from the recovery of unique peptides SV/P2 ratio significantly higher than 2 ( 0.05). We used this of individual proteins. For instance, 116 unique peptides were criterion to define the bona fide “SV-resident” protein group. It identified for the large AZ protein Piccolo whereas the HD method comprised all previously established SV proteins (2, 16, 17), as recovered only 14, and only 1 was identified in the previous study (2) well as hitherto uncharacterized proteins (see The “Hidden SV (Fig. 1B). The UD method increased not only the size of the SV Proteome” Uncovered by the UD Proteomic Method). On the other proteome, but also the number of isoforms identified within indi- hand, a majority of the 1,466 proteins had an SV/P2′ ratio lower vidual protein families, such as the Syts (Fig. 1D), for which most than 1, suggesting that these occasionally interact with SVs. We known family members were detected (13 of 15 and extended-Syt1). defined them as potential SV-visitor proteins. This repertoire The previously undetected isoforms include Syt7, which was recently contains 1) cytosolic proteins, such as , , and found to regulate multiple modes of release synaptojanin-1; 2) AZ proteins, such as Piccolo and Bassoon; (8–10). In contrast, the HD method added only one Syt isoform to and 3) plasma membrane proteins, such as -1, all of the previous SV proteome (2). which interact transiently with SVs, for instance, in the SV As expected, the UD method also detected a much greater trafficking pathway (4, 18, 19). Thus, UD proteomics provide number of proteins (4,439) in synaptosomal fractions (P2′) than quantitative information to distinguish SV-resident and SV- the HD method (1,790) (SI Appendix, Fig. S1A), indicating that visitor synaptic protein repertoires.

Taoufiq et al. PNAS | December 29, 2020 | vol. 117 | no. 52 | 33587 Downloaded by guest on October 1, 2021 A + urea + LysC + Tryp / LysC (off-line) SDS (6h, 37C) (18h, 37C) ERLIC fractionation DTT IAA (on-line) Reverse phase + chromatography

unfolding trimming digesting B C Takamori et al Cell 2006 HD method UD method 1,466 (UD) 766 (HD) 409

Purified synaptosomes Purified synaptic vesicles (kDa) P2’ SV D E (P2’) (SV) F (Takamori 260- et al 2006) (HD) (UD) 160- Syt1 Syt1 Syt1 110- Syt2 Syt2 ext-Syt1 80 - Syt5 Syt5 Syt2 Syt12 Syt12 Syt3 60 - Syt17 Syt17 Syt5 3µm 200 nm Syt-l-4 Syt6 50 - Syt7 Free SVs Syt9 Mitochondria 40 - Syt11 Docked SVs Syt12 30 - Syt13 AZ Syt17 20 - Syt-l-4 PSD Syt-l-5 200 nm

50 nm

Fig. 1. UD proteomics tripled the known SV proteome size. (A) Key steps in the UD proteomics method: sequential enzymatic digestion steps followed by orthogonal peptide separations using multiple biophysical properties of amino acids (SI Appendix, Fig. S1) (IAA, iodoacetamide). (B) Unique peptide coverage by MS of the AZ protein Piccolo (highlighted amino acid sequence) in Takamori et al. (2), HD and UD proteomic methods. (C) Numbers of proteins identified in the SV fraction by Takamori et al. (2) (yellow), HD (white), and UD (gray) methods. (D) Syt family members identified in the SV fraction by Takamori et al. (2), HD and UD methods. (E) EM images of the purified (P2′) and SV fractions, showing clear vesicles with diameters of ∼40 nm. (Bottom) Rep- resentative synaptic structures in the P2′ fraction in EM images, showing intact subsynaptic compartments, as illustrated (PSD, ). (F) SDS/ PAGE profiles of proteins extracted from the P2′ and SV fractions.

We next ranked the 1,466 proteins detected in the SV fraction categories. The first example selected from the inventory is the by iBAQ abundance (Fig. 2C and Dataset S1). We confirmed GTPases, which function in vesicle transport to specific that previously reported canonical transmembrane proteins and subcellular organelles and membranes (20). They are evolu- lipid-anchored proteins were highly abundant (see the word tionally conserved, displaying 75 to 95% amino acid sequence cloud chart in Fig. 2C). The 180 most abundant protein species identity. Such high homology has hampered proteomic detec- accounted for 90% of the total protein mass of SVs, having tion, but, using UD proteomics, we detected and quantified 40 iBAQs of >1.2 × 108 (1.2E8) (Fig. 2C). The iBAQ of the Rabs in the SV fraction, of which 8 were hitherto unreported. Of remaining 1,286 proteins ranged from E5 to E8. Previously, copy 32 Rabs previously documented (2), abundance was quantified numbers per SV were estimated for abundant SV proteins to for only 18 using Western blot analysis (21). We found a majority construct an “average SV” model (2, 6). Using isotope-labeled of high-abundance Rabs (25 of 40) significantly enriched in the SV peptides, we extended the copy number estimate to all other fraction (SI Appendix,Fig.S4B). Among them, Rab11A and detected SV proteins (SI Appendix, Fig. S3 and Table S1). As a Rab11B are highly homologous, with 91% amino acid sequence calibration standard, we utilized the previously determined copy identity (Fig. 3A). Despite such similarity, they reportedly function number of Syt1: 15 (2). The copy number estimated by this in opposing endosomal sorting routes (22). We found 14 unique method for Rab3A was 10.5, which nearly coincided with the peptides common to both Rab11A and -B; however, only UD copy number of 10 previously determined by immunoblotting proteomics could detect a Rab11A signature in the C-terminal (2), confirming the accuracy of this method. These analyses in- hypervariable region. Thus, UD proteomics can reveal highly ho- dicated that copy numbers of many SV proteins are below 1, mologous, but functionally distinct, proteins. suggesting that they are present only in subpopulations of SVs or The second example is the vacuolar-type H+-ATPase (V-ATPase) only transiently interact with SVs. protein complex, which operates as an ATP-driven to energize SVs for neurotransmitter uptake. The V-ATPase complex The “Hidden SV Proteome” Uncovered by the UD Proteomic Method. is composed of a cytoplasmic domain “V1” comprising eight sub- To reveal the hidden SV proteome, we tabulated an SV protein units (A to H), and a transmembrane domain “V0” assembled from inventory detected by UD proteomics with annotations (Dataset four subunits (a, c, d, and e) (23) (Fig. 3B). Previous proteomic S1), including comparisons with those by Takamori et al. (2). studies estimated the copy number of V-ATPase as ∼1to2perSV, This inventory allows one to extract novel insights into SV but the complete set of V-ATPase proteins remains unidentified (2, structure and function using various filters, such as gene family 4, 17). Intriguingly, using UD proteomics, we identified all compo- names, abundance rank, and molecular, structural, or functional nents of the V-ATPase complex, most of which were found in the

33588 | www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Taoufiq et al. Downloaded by guest on October 1, 2021 AB

C NEUROSCIENCE

Fig. 2. Synaptosomal organization and diversity of the SV proteome revealed by UD proteomics . (A) Numbers of synaptosomal proteins detected in either or

both P2′ and SV fractions. (B) Volcano plot showing synaptosomal protein distribution in SV over P2′ fractions; x axis (log2 scale), mean iBAQ (SV/P2′) ratio; y axis (log10 scale), probability of statistical significance (P value, from three independent experiments). The horizontal red dashed line indicates P = 0.05; vertical dashed lines indicate ratios of 1/2 and 2, respectively. (C) Relative abundance of SV proteins in a “word cloud” representation (Inset) and in a ranked (iBAQ) abundance curve of the SV proteome. Shown are estimated copy numbers per SV (SI Appendix, Fig. S3) are indicated with horizontal dashed lines. The vertical dashed line indicates proteins accounting for 90% of the total mass of SV. Names of representative proteins are indicated. Previously undetected SV proteins are marked in red. Abbreviations in parentheses denote the following: TM, transmembrane; LA, lipid-anchored; and S, soluble proteins.

high-abundance range of the SV proteome (Fig. 3B and Dataset S1). substrate (atypical slc subfamily) (30). In addition to these well- Furthermore, V-ATPase accessory proteins Wdr7 and renin recep- known transporters, UD analyses revealed nine SV-resident tor (atp6ap2), and previously hidden Dmxl1 and Dmxl2, were all transporters (Fig. 3C), among which slc10a4 reportedly trans- identified (24) (Fig. 3B). These low-abundance accessory proteins, in ports bile acids into SVs to modulate activity (31). The which only renin receptors are categorized as SV-resident (Dataset remaining eight transporters are orphan slcs of unknown function S1), may regulate V-ATPase complex functions in a restricted subset (SI Appendix, Table S2). Thus, UD proteomics have unveiled and of SVs. Thus, the UD method can reveal full sets of subunits com- quantified hidden transporter proteins of both high and low prising large protein complexes. abundance in the SV proteome, having ubiquitous or restricted The third example is SV-resident transporter proteins. Solute presence in SV populations. carrier (“slc”) transporters are transmembrane proteins that The fourth example is a discovered protein in the SV fraction control movements of soluble molecules across cellular mem- (Fig. 4). In data banks, this protein is known as RGD1305455 branes. To date, more than 400 slc have been identified in (Uniprot ID A0A0G2KAX2) or as “uncharacterized protein C7orf43 mammals, of which ∼40% remain uncharacterized with respect to homolog” and “similar-to-hypothetical protein FLJ10925.” Nothing is their expression profiles and functions (25). Our UD analysis known regarding its tissue expression, developmental profile, or detected slc transporters both in SV-resident and SV-visitor rep- subcellular localization. Six unique peptides from RGD1305455 were ertoires (Fig. 3C). The latter may include transporters partially detected only in UD experiments (Fig. 4A). RGD1305455 was found internalized from the plasma membrane into SVs during endo- as an SV-resident protein (SV/P2′ ratio = 3) of low abundance (rank cytosis (SI Appendix, Fig. S5). SV-resident transporters include 407; copy number/SV ∼0.04) (Fig. 4 A and B). It harbors a conserved VGLUT1 (slc17a7) and VGLUT2 (slc17a6) responsible for glu- DUF domain (DUF4707) and lacks a predicted transmembrane tamate uptake, and VGAT (slc32a1) for GABA and up- domain. Database searches revealed that the protein is highly take, all of which define the molecular identities of the major SV conserved among mammals (>97% amino acid identity) (Fig. 4C populations in the brain (26), and which occur at high abundance and SI Appendix,TableS3). To confirm its presence in the SV in the SV proteome (Fig. 3C). UD proteomics also detected lower fraction, we employed a targeted proteomic strategy. The UD abundance SV-resident transporters that were missing in previous unique peptide VLVVEPVK (Fig. 4A) was chemically synthesized 13 15 SV proteomic studies. These include VMAT2 (slc18a2) (27), using a “heavy” C-terminal lysine ( C6 and N2)andmixedwitha ChT1 (slc5a7) (28), VAChT (29), involved in uptake of mono- digested SV protein sample. A parallel reaction monitoring (PRM) amines or ACh into SV subpopulations, and SVOP of unknown assay based on elution time, ionization, and fragmentation of the

Taoufiq et al. PNAS | December 29, 2020 | vol. 117 | no. 52 | 33589 Downloaded by guest on October 1, 2021 PM1 G1 PM2

PM3 A G2 G3

1.E+11 detected Rab11 peptides specificity method Rab11b NILTEIYR Rab11B HD/UD Lipidation motif NNLSFIETSALDSTNVEEAFKNILTEIYR Rab11B HD/UD 1.E+10 Rab11a NGLSFIETSALDSTNVEAAFQTILTEIYR Rab11A UD only AITSAYYR Rab11A and B HD/UD C-terminal hypervariable extension SIQVDGK Rab11A and B UD only 1.E+9 VVLIGDSGVGK Rab11A and B HD/UD Copy #/SV SIQVDGK Rab11A and B UD only GAVGALLVYDIAK Rab11A and B UD only 1.E+8 1 FTRNEFNLESK Rab11A and B HD/UD 0.04 STIGVEFATR Rab11A and B HD/UD 1.E+7 0.005 HLTYENVER Rab11A and B HD/UD SNLLSR Rab11A and B UD only

Abundance score AQIWDTAGQER Rab11A and B HD/UD 1.E+6 DHADSNIVIMLVGNK Rab11A and B HD/UD NEFNLESK Rab11A and B UD only 90% GTRDDEYDYLFK Rab11A and B HD/UD 1.E+5 AVPTDEAR Rab11A and B UD only %1 200 400 600 800 1000 1200 1400 1600 Rank

B Canonical subunits : Regulatory subunits : 5. V-ATPase V0-a1 (TM) 106. Wdr7 (S) 10. V-ATPase V0-d1 (TM) 121. Renin receptor (TM) 15. V-ATPase V1-A (S) 122. Ac45 (TM) 18. V-ATPase V1-E1 (S) 163. V-ATPase V0-a2 (TM) 5 10 Canonical structure: + Regulators: 15 26. V-ATPase V1-B2 (S) 190. Dmxl2 (S) 18 1.E+11 60. V-ATPase V1-B1 (S) 500. Dmxl1 (S) 26 988. Tm9sf4 (TM) B 69. V-ATPase V1-G1 (S) A A 60 70. V-ATPase V1-C1 (S) 1054. V-ATPase V0-e2 (TM) 1.E+10 69 86. V-ATPase V1-G2 (S) 70 B B 86 87. V-ATPase V1-H (S) E 87 88. V-ATPase V0-c (TM) A Dmxl1 88 101. V-ATPase V1-D (S) G 1.E+9 101 V1 105. V-ATPase V1-F (S) 105 Copy #/SV Tm9sf4 H 1.E+8 1 D C Wdr7 0.04 F (rabconnectin 3-beta) cytoplasm Dmxl2 1.E+7 0.005 d 190 (rabconnectin 3) Abundance score 163 122 106121 V0 ! 1.E+6 500 a c

90% 988 1054 lumen V0e2 1.E+5 Renin %1 200 400 600 800 1000 1200 1400 1600 Ac45 receptor Rank C P2’ SV

3005 1419 47 VGLUT3 Slc36a3 slc12a5 (=KCC2) Mfsd6 VGLUT1 Slc10a4 1.E+11 VGLUT2 Slc45a1 ,CTL2 VGAT (choline) VAT-1 VAT-1 1.E+10 Slc5a7 VMAT2 slc6a11 VGLUT1 Nipal3 (GABA) Slc35g2 Slc5a7 VMAT2 1.E+9 Slc35g2

score Slc22a17 EAAT1 VGAT Slc7a4 ( p -value) VGLUT2 VAChT Lmbrd2 Copy #/SV

(glutamate) ce 10 Nipal3 VGLUT3 GlyT-1 Lmbrd2 1.E+8 1 SVOP -log (glycine) Slc22a17 0.04 Slc7a4

Abundan 1.E+7 0.005 VAChT SVOP 1.E+6 Slc36a3 Slc10a4 90% Mfsd6

0123456 Slc45a1 1.E+5 1/1024 1/32 1321024 1 200 400 600 800 1000 1200 1400 1600 SV/P2’ Rank

Fig. 3. UD proteomics unveiled hidden proteins in both high- and low-abundance ranges of the SV proteome. (A) UD proteomics distinguish highly ho- mologous protein isoforms, Rab11A and Rab11B. (Left) Positions of Rab11A and Rab11B in the ranked (iBAQ) abundance curve. (Middle) Amino acid sequence alignment of Rab11A and Rab11B, showing 91% identity. (Right) List of unique Rab11 peptides detected in the SV fraction by HD and UD methods. Rab11A is identified only by UD from a unique peptide at the C-terminal region of Rab GTPase. (B, Left) V-ATPase–related proteins detected in the SV fraction on the

ranked (iBAQ) abundance curve. Right illustration: Structural model of the V-ATPase protein V0 (a, c, and d) and V1 (A to H) subunits in SVs. Proteins revealed by UD proteomics are indicated in red. (C) SV-resident transporter proteins revealed in the SV fraction by UD proteomics (red), known but missing in previous proteomic studies (purple) in the SV-P2′ volcano plot (Left) and Venn diagram (Right Top). (Right Bottom) Position of the transporters in the ranked (iBAQ) abundance curve of the SV proteome.

33590 | www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Taoufiq et al. Downloaded by guest on October 1, 2021 heavy peptide detected a matched VLVVEPVK peptide in the SV Simultaneous recordings of presynaptic and postsynaptic APs in- sample (Fig. 4D). Close comparison between observed and expec- dicated that the Aak1 inhibitor (1 μM) significantly impaired the ted peptide fragments (SI Appendix,TableS4) indicated that mass fidelity of neurotransmission, assayed as a ratio of postsynaptic errors of native fragments fell within 0.02 dalton (Fig. 4D), con- APs generated in response to presynaptic APs (P < 0.01, n = 6) firming with high precision that protein RGD1305455 indeed exists (Fig. 5I). Altogether, our data indicate that Aak1 is a canonical in the SV fraction. Likewise, in PRM assays using 15 other heavy SV-resident protein with an essential functional role in mainte- peptides for hitherto unidentified SV-resident proteins (SI Appen- nance of neurotransmission, particularly at high frequency. dix,TableS5),thepresenceofallofthetestedproteinsintheSV fraction was confirmed. Many Low-Abundance SV Proteins Are Linked to a Diverse Range of Physiological Functions and Neurological Disorders. Many proteins Functional Characterization of an SV-Associated Protein, Aak1. were uncovered by UD proteomics in both high- and low- The SV fraction contained numerous nontransmembrane pro- abundance ranges of the SV fraction proteome, with >80% teins, some of which reside with SVs within the synaptic com- found in lower ranges (Dataset S1). Even though expressed at low partment (Dataset S1). These proteins might play a regulatory abundance, SV proteins may play important physiological roles. role in neurotransmission. To address this, we focused our We investigated this possibility using functional and disease an- analyses on protein , which are mostly soluble cytoplasmic notations in our database, by classifying SV proteins into 17 proteins. We identified AP2-associated protein kinase 1 (Aak1) functional categories with 26 subcategories (Fig. 6A). Our dataset as an abundant and SV-resident kinase (Fig. 5 A and B). The contains trafficking proteins including SNAREs involved in vari- copy number of Aak1 was calculated as 1.5/SV (SI Appendix, Fig. ous membrane fusions (26 protein species) (2, 6). It also contains S3 and Table S1), suggesting a ubiquitous presence among SVs many types of Rab GTPases (40 species) and membrane-tethering in central synapses (Fig. 5B). The enriched profile of Aak1 in the Trapp complexes (14 proteins) (SI Appendix,Fig.S4). Many of purified SV fraction was confirmed by Western blot, contrasting these proteins are identified as SV-resident, suggesting that SVs with other cytoplasmic kinases found in P2′, such as MARK2 or may be equipped with proteins for various trafficking routes to- TNiK (SI Appendix, Figs. S2E and S6B). In cultured hippocam- ward other presynaptic organelles. Other major categories in- pal , strong colocalization of exogenously expressed cluded proteins involved in signaling (e.g., kinases, phosphatases), Aak1 (TagRFP-Aak1) with an SV marker, - signal transduction, and transport of small molecules. UD pHluorin (SypHy), was observed (Fig. 5C). proteomics detected a high number of metabolic enzymes (179 We employed both genetic and pharmacological approaches species), including those involved in neurotransmitter metab-

to clarify the functional role of Aak1, using short hairpin RNA olism (13 species), cellular energy production (35 species), lipid NEUROSCIENCE (shRNA) knockdown (KD) of Aak1 expression in cultured hip- regulation (75 species), and cyclic nucleotide second messen- pocampal neurons, and by infusing an Aak1-specific inhibitor, gers (12 species). These data suggest the occurrence of meta- LP-935509 (32), directly into the calyx of Held presynaptic ter- bolic reactions on SVs in crowded presynaptic terminals (6). minals in brainstem slices of rats at postnatal day (P) 13 to 15. UD proteomics also detected SV proteins categorized as For Aak1-KD, we applied a lentivirus targeting Aak1 at day 11 -related proteins (40 protein species) (SI Appendix, in vitro (DIV11), when synaptophysin became detectable in Fig. S4D). The presence of both SV-resident (e.g., , Western blot (SI Appendix, Fig. S6A). At DIV15, the KD effect atg9a, trappc8, and pik3c3) and SV-visitor autophagy-related became maximal, reducing Aak1 expression below 5% (Fig. 5D). proteins (e.g., beclin-1, uvrag, map1lc3a, and cisd2) in our SV In hippocampal culture at DIV15, excitatory postsynaptic cur- proteome (SI Appendix, Fig. S4D and Dataset S1) suggests that rents (EPSCs) in Aak1-KD neurons underwent a rapid short- autophagic degradation may participate in the maintenance of SV term depression (STD) during stimulation at 20 Hz. The mag- population size within presynaptic terminals. nitude of STD was significantly greater than that in controls To examine the pathological implications of our UD proteo- (P < 0.05, n = 7) (Fig. 5E). Consistently, at the calyx of Held mics data, we searched for genetic information on SV proteins loaded with LP-935509, EPSCs underwent stronger STD during regarding their associations with neurological diseases (see a 100-Hz train compared to controls (0.3 s, P < 0.05, n = 7) (SI “Diseases in the SV fraction” in Dataset S1) and marked them in Appendix, Fig. S6C). Cumulative histograms of EPSC amplitudes ranked abundance plots of SV (Fig. 6B) and P2′ fraction pro- provided the pool size of readily releasable SVs and release teomes (SI Appendix, Fig. S7). We found that 236 different brain probability, indicating that both Aak1-KD (Fig. 5E) and Aak1 diseases are associated with 210 high- and low-abundance pro- inhibitor (SI Appendix, Fig. S6C) reduced the pool size without teins of the SV proteome, of which 159 (76%) are revealed by affecting the release probability. Furthermore, the recovery from the UD method. Likewise, 55% of these SV proteins were found STD was prolonged, both at the Ca2+-dependent fast component in low-abundance ranges of the P2′ proteome (SI Appendix, Fig. (33) and Ca2+-independent slow component at the calyx of Held S7). These results indicate the pathological significance of SV (Fig. 5F). These results together suggest that Aak1 normally proteins irrespective of their abundance. SV protein-associated facilitates SV recycling, thereby maintaining the releasable SV diseases include many motor (145 associated proteins), cognitive pool. (135 proteins), and sensory system phenotypes, such as visual (33 To further investigate whether Aak1 is involved in exo- proteins) and auditory (14 proteins) phenotypes. The database also of SVs, we performed pHluorin assays in cultured indicates SV proteins associated with phenocopy diseases, such as hippocampal neurons (Fig. 5G) and capacitance measure- mental retardation (28 disease phenotypes), epilepsy (25 pheno- ments at the calyceal terminal (Fig. 5H). In pHluorin assays, types), Parkinson’s disease (13 phenotypes), amyotrophic lateral endocytic fluorescence half-decay time was prolonged by twofold sclerosis (4 phenotypes), Alzheimer’s disease (4 phenotypes), and (P < 0.005, n = 51) compared to controls (n = 20). Likewise, in cerebellar ataxia (10 phenotypes). Our UD cross-analyses between capacitance measurements, LP-935509 (1 or 10 μM) significantly functions and diseases indicate that phenocopies may involve pro- prolonged the endocytic capacitance change. Capacitance mea- teins from both SV-resident and SV-visitor repertoires, from surements did not indicate a significant reduction of exocytosis. both high- and low-abundance ranges, and from functionally Thus, both at hippocampal and brainstem synapses, Aak1 likely distinct proteins in the SV life cycle. For example, Parkinson’s plays an accelerating role in SV endocytosis. disease can be linked to in SV-resident proteins such Since the above results suggest involvements of Aak1 in the SV as renin-receptor (121st rank), involved in SV acidification, recycling pathway, we further investigated whether Aak1 might dnajc13 (318th rank, SV endocytosis) and sv2c (97th rank, SV have a physiological role in the maintenance of neurotransmission. trafficking), or in SV-visitor proteins, such as synaptojanin-1

Taoufiq et al. PNAS | December 29, 2020 | vol. 117 | no. 52 | 33591 Downloaded by guest on October 1, 2021 AB

RGD1305455 456

1.E+11

1.E+10 ( p -value) 10 RGD1305455 -log10(adj p.value) 1.E+09 Copy #/SV -log 1.E+08 1

0.04 123 1.E+07 0.005 Abundance score

1.E+06 0 90% 1/1024 1/32 1321024 1.E+05 10 200 400 600 800 1000 1200 1400 1600 SV/P2’ Rank

C B. floridae 86 HHEGGLERGTVPVSNGKQDFRDCVPLLLHNSSAKDQAE------KGEDPVTV P. humilis 126 -PGNFG-NDEREDSKEPPVFRECRALLTHSRGPPGSAGA------GLPVDDPIV A. ocellaris 140 AAEAAIAALGSRVDSRCRNFRDCKPLLIHNSSGT------ASREFRRAPVQSPLDEPVV P. sinensis 87 PPEGPG-TDGAAPPPNPSQFRDCCPLLTHGQGPPGRPAA-----GVRRDPGEIPVEEPIV Delphinapterus Loxondonta Callithrix Pongo Homo D. leucas 297 RPEGRGTLEGSGRPMTGEERGGREPHTADLKGNAGAAGGLDSSDLFTLFTTQLPVEEPIV R. norvegicus 109 DPGGGGLFRS------CSPLLTHGPG------PATSGGATTLPMEEPIV leucas africana jacchus abelii sapiens M. musculus 109 DPGGGGLFRG------CSPLLTHGQG------PATSGGATTLPVEEPIV L. africana 289 DPGGGGLFRG------CSPLLTHGPG------PATSGGATTLPVEEPIV P. abelii 110 DPGGGGLFRG------CSPLLTHGPG------PATSGGATTLPVEEPIV H. sapiens 109 DPGGGGLFRG------CSPLLTHGPG------PATSGGATTLPVEEPIV C. jacchus 109 DPGGGGLFRG------CSPLLTHGPG------PATSGGATTLPVEEPIV

B. floridae 179 ------PAESV------EEQVT--KEVKESSDVKCHV------Y P. humilis 292 GTGGYLRLLQGRAPGQVFRQQHGAFKAQVSTLLTVLPPPRVRCRQVTVSGKYLTVLKVLN A. ocellaris 234 --LGYLSVLQQREPTHTFRHDLNTFKAQVSTTLTVLPSPTVRCKQMTVSGRQLAVLKVLN P. sinensis 182 --HGYLSLLQNRAPGQLFHEEQGAFKAQVSTMLTVLPPPGLKCRQLNVSGKYPTXXSVLN (23%) (0%) (0%) (0%) (0%)%%) D. leucas 398 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLKCRQFTVAGKHLTVLKVLN R. norvegicus 187 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLKCRQFTVAGKHLTVLKVLN (99%) (97%) (98%) (97%) (98%) M. musculus 187 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLKCRQFTVAGKHLTVLKVLN L. africana 367 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLKCRQFTVAGKHLTVLKVLN 99% 97% 97% 97% 97% P. abelii 188 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLRCRQFTVAGKHLTVLKVLN 791 760 580 581 580 H. sapiens 187 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLRCRQFTVAGKHLTVLKVLN C. jacchus 187 --QGYLRLLQTRSPGETFRGEQSAFKAQVSTLLTLLPPPVLRCRQFTVAGKHLTVLKVLN (beluga (African savana (white-tufted- (Sumatran (human) whale) elephant) ear marmoset) orangutan) B. floridae 203 NSSCEDVTVHQLQIIPNCNAAFQHDSAEKTSPQPNIQTPSKRTAEGGLFSLVHTDPSQLP P. humilis 352 GSSQEELSLWDVQILPNFNASYLPVMPDGSVLLVDDVCHHSGEVPVGAFCRVPGGQAGWP A. ocellaris 292 ESSQEEVSIRDVRILPNLNASYLPMMPDGSVLLVDNVCHQSGEVGMASFCRVDSLACRLP P. sinensis 240 GCSQDEISLWDIRILPNFNASYLPMLPDGSVMLVDDVCHHSGEVPVGAFCRVAGAGSSCP D. leucas 456 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGCFP Pelodiscus Pseudopodoces Amphiprion Branchiostoma R. norvegicus 245 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGYFP M. musculus 245 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGYFP sinensis humilis ocellaris floridae L. africana 425 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGCFP P. abelii 246 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGCFP H. sapiens 245 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGCFP C. jacchus 245 SSSQEEISIWDIRILPNFNASYLPVMPDGSVLLVDNVCHQSGEVSMGSFCRLPGTSGCFP gap B. floridae 263 SHLQPLEELCLVFQIVVNDNWNAGKYESHESALLLLMLWQPGKAVNSNSSEEHIHTHYSL P. humilis 412 CPLSALEEQNFLFQLQAPERPPCDSKEGLEVPLVAVVRWSTPKLPF----TNSIVTHYRL (4%) (3%) (9%) (13%) homology A. ocellaris 352 SMLSSLEEHDFLFQLHLNDMPQDDSNEGLEVPLVAVLQWSTPKMPF----TNCIYTHYRL P. sinensis 300 CALSALEEQNFLFQLQAPERPQEDTKEGLEVPLVAVVQWSTPKLPF----TSSIYTHYRL (80%) (86%) (70%) (47%) D. leucas 516 CLLSALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL 73% 77% 61% 33% identity R. norvegicus 305 CPLSALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL M. musculus 305 CPLSALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL 575 674 618 557 L. africana 485 CPLSALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL P. abelii 306 CPLNALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL (Chinese soft- (Tibetan (clown (Florida H. sapiens 305 CPLNALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL C. jacchus 305 CPLNALEEHNFLFQLRGGEQPPPGAKEGLEVPLIAVVQWSTPKLPF----TQSIYTHYRL shelled turtle) ground-tit) anemonefish) lancelet)

D (Heavy peptide) (Native peptide) Intensity (%) y7 y6 y5 y4 y3 y2 y1 Intensity (%) y7 y6 y5 y4 y3 y2 y1 V L V V E P V k V L V V E P V K 100 y3 2b b3 100 2b b3 y3 y6 50 b2 50 y6 y5 b2 3 y1 y4-H2O y4-NH y1 y5 y4 y7 y4-H20 y1-H2O y2 y2 b3 y4 y7 m/z m/z 100 200 300 400 500 600 700 800 100 200 300 400 500 600 700 800 900 1000

VL V VEPVk bMax VL V VEPVK bMax k V P EVVL V yMax KVPE VVL V yMax

Mass error (Da) Mass error (Da) 0.02 0.02 0.0 m/z 0.0 m/z -0.02 -0.02

Fig. 4. A previously hidden SV-resident protein shows high amino acid among mammals. (A) “Uncharacterized Protein RGD1305455” (Uniprot ID A0A0G2KAX2), identified in UD proteomics from unique peptides (highlighted within amino acid sequence) and its position in the ranked (iBAQ) abundance plot (Lower). No unique peptide could be detected with the HD method. (B) SV-resident position of RGD1305455 in the SV-P2′ volcano plot. (C) Amino acid sequence comparison of RGD1305455 homologs in various animal species. Black and gray shading indicates identical and similar amino acids, respectively. Dashes represent gaps in sequences. See SI Appendix, Table S4 for protein accession numbers and reference sequences. Mammal species are framed in dashed red line boxes (Left) and species pictures with >97% identity (Right). (D) Confirmation of the existence of RGD1305455 protein in the SV 13 15 fraction using a targeted proteomic approach. VLVVEPVK peptide (detected in UD proteomics) was synthesized using C-terminal “heavier” [ C6 N2] lysine (+ eight neutrons = a predefined shift of 8 Da) and was used to track native peptides after mixing with digested SV proteins. (Left) MS2 spectra of heavy VLVVEPVK peptide. (Right) MS2 spectra of a native peptide detected in the SV fraction that coincides with that of the heavy peptide. Red, blue, and black peaks indicate matched y-ion series, b-ion series, and unmatched ions respectively (Upper). (Middle) Amino acid sequences corresponding to the ion frag- ments. (Lower) Plotted mass errors of detected versus expected peptide fragments. Errors of native peptide fragments were all <0.02 dalton. Expected masses of all fragments are specified in SI Appendix, Table S5.

33592 | www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Taoufiq et al. Downloaded by guest on October 1, 2021 (351st rank, SV endocytosis) or pla2g6 (653rd rank, lipid It needs to be borne in mind that SVs, starting from enriched composition) (Dataset S1). Altogether, our data analyses il- synaptosomes, are isolated solely based on their size and density. lustrate the complexity and physiological importance of low- Therefore, heterogeneity may also be caused, at least in part, by abundance SV protein repertoires revealed by UD proteomics. the presence of membranes derived from different trafficking steps, such as partially -uncoated vesicles, small endo- Discussion somal vesicles, or SVs from axonal compartments en route to In this study, we have used SVs purified from rodent brain as a nerve terminals. While these compartments are part of the same model for identifying and quantifying the “deep proteome,” recycling pathway and are expected to share vesicular applying our proteomic workflow. SVs isolated from mammalian membrane-resident proteins, the “visitor” proteins are likely to brain are morphologically homogeneous (34) and share a set of be different. This may explain the presence of endosomal-related common proteins, with more than 90% containing the major SV proteins (e.g., Stx7, AP3) or proteins of the AZ (e.g., Piccolo, protein synaptophysin (2). Yet, they are heterogeneous with re- Bassoon) in the SV proteome. Moreover, we could not exclude spect to synapse types and neurotransmitter content. With the the possibility that the SV preparation is contaminated, even to a proteomic workflow introduced here, we identified ∼1,500 pro- small extent, with vesicles from other sources: for instance, small teins in SVs, more than three times as many as previously reported vesicles artificially generated from larger membranes during (2, 4, 7). Of these, we found 134 SV-resident proteins, of which 86 homogenization, or vesicles from the postsynaptic side. Indeed, are of low abundance (<1 copy per SV). These proteins may analysis of the UD-SV proteome using the SynGO resource (15) therefore be restricted to SV subsets, deduced from the findings has revealed a postsynaptic contamination of at least 4% based that they include previously missed vesicular transporters for on 691proteins that were annotated in SynGO. Regarding pos- monoamines and acetylcholine, present in only a small percentage sible contaminants in the remaining 775 SV proteins in our of brain synapses. Of the ∼1,500 SV-fraction proteins, more than study, we cannot make a definitive calculation as these are not 200 have genetic associations with CNS diseases, highlighting the annotated in the SynGO database. importance of this deep diverse and previously hidden proteome A closer look at the defined SV-resident repertoire (proteins ′ > for proper brain functions. A resource database (Dataset S1)was with SV/P2 iBAQ ratios of 2) (Dataset S1) provides important constructed to include all data on identification, quantitative dis- leads toward a better understanding of SV molecular and func- tional heterogeneity. Of 134 SV-resident proteins, 86 have copy tribution, and structural and functional annotations for each < SI Appendix protein detected in the SV fraction. numbers 1/SV ( , Fig. S8). The 40 most abundant SV- The increased peptide coverage of the “UD workflow” is resident proteins (in ranks 1 to 180) include all of the subunits of “ ” NEUROSCIENCE based on two major improvements in combination: 1) enhanced V-ATPase, vesicular tetraspanins including SCAMPs, synapto- cleavage using in sequence and 2) the introduction of physins, and synaptogyrins, Syts and SV2 proteins, as well as an off-line orthogonal peptide separation prior to reversed-phase membrane-associated and CSPs. VGLUT1/2 and liquid chromatography tandem mass spectrometry (LC-MS/MS). VGAT, vesicular transporters of the two major neurotransmitters in the brain, glutamate (excitatory synapses) and γ-aminobutyric These steps resulted in a remarkable increase in unique peptide acid (GABA) (inhibitory synapses), are also in this list. All these detection and have greatly expanded the protein inventory of proteins are likely present on SVs throughout the entire nervous SVs, including highly homologous proteins within families. For system. example, 40 Rab proteins, having high sequence homology (75 to Minor SV residents (<1 copy per SV, beyond rank 180) in- 95%), but distinct trafficking functions (20), were identified. clude proteins generally involved in membrane trafficking, such Likewise, functionally characterized but hidden Syts, such as Syt7 – as additional SNAREs, Rab GTPases, kinases, (8 10), were detected, together with other family members of tethering complexes, and autophagy-related proteins. Their low unknown functions. Moreover, the high peptide yield of UD abundance suggests that they reside on a subset of vesicles within proteomics allows unprecedented label-free and highly reli- synapses. For example, the copy number of the transmembrane able quantification of most proteins in the dataset. We were protein Atg9a was 1 per 25 SVs (SI Appendix, Fig. S3 and Table able to evaluate the copy numbers of many hundreds of pro- S1), implying that 4% of vesicles in the synaptic compartment teins, thereby providing a quantitative scope of the whole SV may be recruited to an autophagic pool. As another possibility, proteome organization. When compared to previous quanti- these proteins may be expressed specifically in a small subset of tative studies (2, 6), the results largely confirm copy numbers synapses in specific brain regions. Indeed, this list includes the — on average per vesicle, except for three proteins SNAP29, known scarce neurotransmitter vesicular transporters VMAT2, — vti1a, and ClC3 that have abundance scores too low to be VAChT, Slc5a7, and VGLUT3, reflecting the functional hetero- further considered as major SV proteins. On the other hand, geneity of synapses (Fig. 3C and SI Appendix, Table S2). Inter- most detected proteins had copy numbers less than 1 per SV on estingly, our list also includes almost a dozen of hitherto unreported average, revealing much greater SV heterogeneity than transporter proteins. previously envisaged. Many SV proteins, whether classified as residents or potential Our label-free quantification also allowed a quantitative visitors, may have specific functions in regulating or maintaining comparison of the proteomes of isolated nerve terminals and the performance of synapses. In fact, our UD proteomics have purified SVs. This was not only the foundation for identifying detected over 200 proteins in the SV fraction known to be ge- bona fide SV residents, but also for distinguishing between SV netically associated with neurological (mental, motor, and sen- resident and potential SV visitor proteins. Remarkably, about sory processing) disorders. Remarkably, a majority of these 50% of the SV residents are nontransmembrane proteins (Dataset proteins (76%) were found in low-abundance ranges and had S1), highlighting the high degree of proteome organization, de- copy numbers of <0.04/SV. These neurological disorders likely spite molecular crowding at the synapse (6). For example, UD originate from various synaptic dysfunctions specific to discrete proteomics revealed that, among nontransmembrane proteins, neuronal populations of the nervous system. In fact, recent evi- Aak1 is a major SV-resident protein, having an SV/P2′ ratio of ∼4 dence supports the idea of “synaptopathies” as a causal poly- and a copy number/SV of 1.5. Our functional assays indicated that genic mechanism for psychiatric diseases (35–38). In the process this kinase is essential to maintain high-frequency neurotransmis- of evolution, abundant canonical proteins are often ancestral sion by accelerating SV recycling. Thus, our classification of SV components whereas proteins of low abundance tend to emerge protein repertoires may facilitate functional studies and may result for new functions (39). In this respect, the deep diversified syn- in the identification of major regulators of synaptic transmission. aptic proteome may account for mammalian- or human-specific

Taoufiq et al. PNAS | December 29, 2020 | vol. 117 | no. 52 | 33593 Downloaded by guest on October 1, 2021 Prkar2a Rock2 Ak1 Csnk2b A Mark1 B 1.E+11 Stk39 Aak1 Pip5k1c Camk2a Csnk2a1 Prkcg Camk2a Mapk1 Camk2d Camk2b 1.E+10 Camk2d Mapk10 Camk2g Prkcb Mapk3 Camk2b Lyn Kit Prkar2b Camk2g Cdc42bpb Mtor Mapk10 Pi4ka 1.E+9 Prkca Csnk1g3 Abl1 Pip4k2b Map2k1 Prkcg Dclk1 Cdk18 Map2k2 Mink1 Yes1 Map2k1 Mapk1 Copy # / SV Mark4 Prkaca Aak1 Csnk1d Cdk17 Pip5k1c Pip4k2aTaok1 Prkar2b Prkacb 1.E+8 Gsk3a Cdk5 Prkar2a 1 Slk Csnk1e Fyn Cdk16 Cdc42bpa Dclk1 Mark2 Pik3c3 Spry2 0.04 Map2k2 Pik3c3 Pi4k2a Pi4ka Slk ( p -value) Nek7 Src Ak1 10 Cdk5 1.E+7 0.005

-log10(adj p.value) Csnk1g1 Cdc42bpa Prkce Brsk2

Pik3r4 Abundance score Nek7 -log Pkn1 Gsk3b Mark2 Mark3 Prkacb Stk39 Pi4k2a Csnk1e Cdc42bpb Rock2 Mark3 1.E+6 Pkn1 Src Prkg2 Prkcd Pik3r4 Prkaca Map2k7 Mapk3

90% Map4k4 Csnk2a2 Prkg2 Csnk2a1 Brsk1 Pip4k2a Dclk2 Mark1 Csnk2b Mark4 Mink1 Kit Pip4k2b Taok1 Dclk1 Prkce 1.E+5

0123456 Yes1 Brsk1 Prkcd %1 200 400 600 800 1000 1200 1400 1600 Gsk3b Map2k7 Map4k41/1024Fyn 1/32 1Prkcb 32 1024 Rank Mtor Csnk1d Prkca Cdk18 SV/P2’ Brsk2 rol Abl1 Csnk1g3 Cdk17 t Spry2 Cdk16 Lyn Csnk1g1 D SypHy SypHy TagRFP Merge uninfectshRNA-coned shRNA-Aak1 C TagRFP 1 Aak1 0.8 0.6 GFP 0.4 (kDa) 0.2 2 m 260 - (normalized) Fluorescence 0 160 - 012345 110 - Distance ( m) 80 - SypHy TagRFP-Aak1 Merge SypHy 60 - TagRFP-Aak1 1 50 - 0.8 0.6 40 - 0.4 2 m 0.2 30 - (normalized) (total protein staining) Fluorescence 0 012345 Distance ( m) 20 - p = 0.005

CSPE 8 4 )A

shRNA-control s) E n F 3 ( 6

evital Conditioning train Test e

d ISI 2

u 4 til

slow ( umuC

pma 1 Aak1-KD 2 100 Hz, 30 times 100 0 9 300 pA 0 )%(yr 0.0 0.2 0.4 0.6 0.8 1.0

200 ms Time (s) 80 )An(edut

Control

e

yt

)An,

)%(C

v

i

lib o 100 1.5 4 0.6 p = 0.6 60 Control LP93550 shRNA-control (n = 7) c LP935509

p = 0.035 p = 0.036 e ab 80 rCSP p = 0.021 Aak1-KD (n = 7) qN 0.1 s 0.3

ilp 3 SP

orpesaele 40

1.0 (e 2s 1 nA

60 s)

Ed 2ms ma 20 s

zi 2 0.3 0.2

E 20 40 s

ezi 0.5 1st EPSC

C PRR fast (

l 1 0.1 S

a 20

PE 0 R

mr 0.0 0 0.0 0 5 10 15 20 25 30

t lor

l 0.0 l

oN 0

DK- D

s1

DK-1kaA o ortn Inter-stimulus interval (s)

0 5 10 15 20 r

K-1kaA

t t

no noC

Number of stimuli 1

oC

k

C aA Control

stimulation LP935509 (10 Hz, 10s) GH1.0 0.6 Control I LP935509 (1 uM) Control LP935509 0.8 control 0.4 LP935509 (10 uM)

ed ΔF 1-2 s Aak1-KD pF) z 0.6 0.2

0.4 Cm ( 21-22 s

Normali 0.2 0.0 41-42 s 0.0 -0.2 20 mV 04080120160 -10 0 10 20 30 40 50 60 Time (s) 200 ms Time (s) Control (n = 6) p < 0.005 p < 0.005 LP935509 (n = 6)

) 40

30 0.6 p = 0.059 50 p = 0.311 ) (s p = 0.021 p = 0.007 100 25 40 )

ime 30 time (s) 0.4 20 pF) pF

ay 30 ay t c

15 c 20

de 50 Cm ( 20 f 0.2 QCa (

10 f de 10 10 Hal 5 Hal 0 0 0.0 0 Output / Input (% 0

D 0 20 40

K Time (s) Control Control Control ak1- Control 1 uM LP 1 uM LP 1 uM LP A 10 uM LP 10 uM LP 10 uM LP

Fig. 5. Aak1 is an SV-attached protein kinase essential for high-frequency neurotransmission. (A) SV/P2′ volcano plot of 69 kinases detected in the SV fraction by UD proteomics. Labels indicate names of genes encoding proteins. Aak1 (circled in red) is one of the few identified SV-residents. (B) SV kinases in the abundance curve. Aak1 is the most abundant of SV kinases. (C) Colocalization of exogenously expressed Aak1 and synaptophysin (SypHy) in cultured hippocampal neurons. (Upper) Control TagRFP (red) and SypHy (green) images and their line-scanned profiles (Right). (Lower) TagRFP-Aak1 (red) and SypHy (green) images and their line-scan profiles showing complete colocalization. (D) Lentivirus KD efficiency of Aak1 confirmed at DIV15 by Western blot analysis. Hippocampal cells were infected at DIV 11 to 12 with lentivirus coexpressing GFP and shRNA targeting Aak1. (E) Aak1-KD enhanced STD of EPSCs and reduced estimated readily releasable pool (RRP) size of SVs (Nq) evoked by a 20-Hz train of stimulation in Aak1-KD (red, n = 7) or shRNA control (black, n = 7) neurons. (F)LP935509(1μM) slowed both fast and slow components of recovery from STD (100 Hz) at the calyx of Held (red traces, n = 6, stimulation protocol indicated on the Top). EPSCs at different interstimulus intervals (ISIs) are shown in Insets.(G) Aak1-KD slowed endocytic SV fluorescence in response to stimulation in pHluorin analyses in hippocampal neurons. Aak1-KD (red), n = 16 neurons, 51 boutons; and control (black), n = 10 neurons, and 20 boutons. (H)LP935509(1or10μM) prolonged the SV endocytic 2+ half-time (P < 0.001, n = 6at1μMandn = 4at10μM: 4; one-way ANOVA: F(2, 13) = 16.81] without affecting exocytic [P = 0.059, F(2, 13) = 2.53] or presynaptic Ca current magnitudes [P = 0.511, F(2, 13) = 0.15] in membrane capacitance measurements at the calyx of Held. (I)LP935509(1μM) impaired fidelity of glutamatergic neurotransmission at 100-Hz at the calyx of Held. (Top) Postsynaptic APs evoked by presynaptic APs in the presence (red) or absence (black) of LP935509 in presynaptic terminals. (Bottom) Percentages of postsynaptic APs elicited by presynaptic APs, controls (n = 6) and the Aak1 inhibitor (n = 6).

neurological diseases. This could be a key reason why, despite Materials and Methods technical difficulties, investigations of deep subcellular proteomes All animal experiments were performed in accordance with guidelines of beyond “average models” are necessary. the Physiological Society of Japan, the German Animal Welfare Act, and

33594 | www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Taoufiq et al. Downloaded by guest on October 1, 2021 A Total SV fraction SV-resident

B Neurological disease category: 1 cognitive 2 motor 6 sensory 1.E+11 9 NEUROSCIENCE 12 16 983 1190 121 1193 969 985 1194 139 384 385 961 993 1.E+10 1001 1197 149 383 386 849 1208 160 653 862 1002 1209 380 387 872 55 163 521 523 658 1003 1237 298 390 659 1004 56 168 301 518 527 877 1238 514 533 661 1027 1.E+9 169 302 888 1036 1253 513 545 669 1254 57 189 306 923 1049 190 309 398 550 680 1256 61 928 1051 191 312 418 556 697 1258 62 210 419 936 1053 318 558 708 1083 1263 213 426 937 1.E+8 319 561 711 1090 1264 939 74 321 437 566 748 1101 1270 326 438 572 940 1114 1271 90 441 751 579 948 1117 94 1292 949 1121 1302 1.E+7 288 953 1147 1308 97 954 1155 1310 275 463 1168 Abundance score 100 1159 171 264 453 1172 1314 104 635 1362 1317 107 178 254 367 512 813 1177 188 252 625 1185 1382 1319 1.E+6 351 499 809 1385 222 241 347 617 839 1321 492 606 785 825 (NA)1140 1386 1330 227 240 343 823 233 336 602 782 1137 1390 1350 239 478 (NA) 817 1394 600 776 814 1133 1354 1128 1395 1361 1.E+5 1399 1% 1400 200 400 600 800 1000 1200 1400 1600 1420 Rank

Fig. 6. Diversity of neuronal functions and dysfunctions related to the UD-SV proteome. (A) Functional mosaic of the SV proteome. Each protein detected in the SV fraction by UD proteomics was associated with one or more functional keywords. (Top Left) Sunburst diagrams show distributions of the functional categories (inner circle) and subcategories (outer circle) represented in the total SV fraction proteome and in the SV-resident repertoire. Shown is a list of functional keywords (Left) and subcategories (Right) with number of proteins representing each category (for details of protein functional annotations, see Dataset S1). (B) The deep low-abundant SV proteome is related to brain diseases. Proteins detected in the SV fraction having “disease(s) caused by (s) affecting the gene represented in the entry” were marked, and their rank in the iBAQ-abundance curve is specified. The analysis was performed manually using the Uniprot and GeneCards databases for human diseases. Markers indicate proteins associated with cognitive (purple), motor (red), and/or sensory processing (yellow) disabilities. The vertical dashed line indicates rank 409, the number of proteins identified by a previous SV proteomics study [Takamori et al. (2)]. Proteins to the right hand of the dashed line were mostly revealed by UD proteomics (see Dataset S1 for a listing of all disease names and protein associations).

regulations at the Okinawa Institute of Science and Technology, at the Max- (Sigma). After shaking 1 min at 850 rpm at room temperature (Eppendorf Planck Institute for Biophysical Chemistry, and at Doshisha University. Thermomixer), samples were centrifuged during 13 min at 6,400 × g (con- ditions that were optimal to remove all of the liquid from the upper Brain Synaptosomes and SV Purifications. Synaptosomes (P2′) and SVs were chamber using a TOMY Kintaro KT-24 centrifuge). After repeating these – purified from whole brain of 4- to 6-wk-old Sprague Dawley rats following steps two times, proteins were resuspended with 200 μL of urea buffer the same protocols used in ref. 2 and previously described in ref. 34 for SV, containing 50 mM iodoacetamide and incubated in darkness for 1 h at room and in ref. 4 for P2′. The quality of all P2′ and SV purification procedures was temperature. The alkylation was then stopped by centrifuging as above and controlled by Western blots of synaptic protein markers and by EM. Extended by resuspending protein samples with 200 μL of urea buffer containing 25 descriptions of biochemical, imaging, and electrophysiological procedures and mM dithiothreitol. Unfolded proteins were subsequently washed with analyses are provided in SI Appendix, SI Materials and Methods. 20 mM ammonium bicarbonate. Proteolytic enzymes were used in a ratio of “ ” UD Proteomics Sample Preparation and MS. 1:50 with proteins. To generate peptides, a first digestion step ( trimming ) Sequential protein digestion. Fifty micrograms of proteins extracted from P2′ or was performed using endoproteinase lys-C (Promega) for 6 h at 37 °C, fol- SV were resuspended into 200 μL of buffer containing 8 M urea, 100 mM lowed by a second digestion step overnight (16 to 18 h) at 37 °C using a Tris·HCl, pH 8 (“urea buffer”) and placed onto a Pall Nanosep Omega filter trypsin/lys-C combination (Promega). After centrifugation as above, digested

Taoufiq et al. PNAS | December 29, 2020 | vol. 117 | no. 52 | 33595 Downloaded by guest on October 1, 2021 peptides were acidified with 1% trifluoroacetic acid, concentrated, and used for peptide separation. Temperature of the heated capillary was dried using an EZ-2 Elite evaporator (SP Scientific). 300 °C, and a 1.9-kV spray voltage was applied to all samples. The mass Orthogonal peptide separations. To separate peptides, off-line ERLIC or ERLIC- spectrometer settings were as follow: full MS scan range 350 to 1,500 m/z based separation was performed (14). The following conditions were with a mass resolution of 70,000, 30-μs scan time, and automatic gain control adapted and optimized to obtain the highest number of identified proteins set to 1.0E6 ions, and fragmentation MS2 of the 20 most intense ions. from P2′ and SV samples. Mobile phase solvent preparation was as follows: Protein identification. Protein identification was done using Proteome Dis- Solvents were freshly prepared for each experiment using liquid chroma- coverer software v2.1 (Thermo Scientific) and Mascot 2.6 (Matrix Science) as a search tography/mass spectrometry grade acetonitrile (ACN), formic acid (FA), and engine. A database downloaded from UniprotKB Rattus norvegicus (proteome ID water from Thermo Fisher Chemicals. Solvent A was prepared as follows: 90% UP000002494) was used with search parameters as follows: trypsin , up to ACN, 0.1% FA. Ammonium hydroxide (NH4OH, 25% weight/weight [wt/wt] in two miscleavages, with precursor and fragment mass tolerance set to 10 parts per water; Fluka) was then added to adjust pH at 4.5. Solvent B was prepared as million and 0.02 Da, respectively. Cysteine carbamidomethylation, methionine follows: 30% ACN, 0.1% FA. The digested peptide mixture was resuspended oxidation, asparagine and glutamine deamidation, and N-terminal protein with 20 μL of solvent A and injected into a weak anion exchange PolyWax acetylation were set as variable modifications. The results were filtered using a column (1-mm inner diameter × 150 mm, 5-mm particle size, 300-Å pore size; false discovery rate of <1% as a cutoff threshold, determined by the Percolator PolyLC Inc.) using a PAL autosampler (CTC Analytics) for automatic injection algorithm in Proteome Discoverer software. and fractions collection, using a gradient mode (3 min solvent A, to 10% B in Quantitative proteomic data statistical analyses. For the volcano plot, iBAQ data 7min,10%Bto25%Bin24min,25%Bto70%Bin16min,70%Bto81%B obtained from Proteome Discoverer were used for statistical analysis using R in6min,81%Bto100%Bin3min,withfinalwashat100%Bfor6minand software version 3.2.5 (R Project for Statistical Computing). Quasi-Poisson reequilibration at 100% A for 20 min) at a flow rate of 40 μL/min. Twenty-four generalized linear models were generated (y ∼1, y ∼ treat) and compared fractions were collected every 3 min between 0 and 72 min and subsequently using analysis of deviance for generalized linear model fits (Anova) to obtain P concentrated to dryness using a speed vacuum Genevac EZ-2 Elite (SP values, using an F-test, and adjusted with the Benjamini–Hochberg method. Scientific). μ MS. Dried peptides were resuspended in 30 L of 0.1% FA and analyzed using Data Availability. Proteomic raw data files data have been deposited in the a Q-Exactive Plus Orbitrap hybrid mass spectrometer (Thermo Scientific) Japan Proteome Standard Repository Database (accession no. JPST000968). equipped with an Ultimate 3000 nano-high-pressure liquid chromatography All other study data are included in the article, Dataset S1,andSI Appendix. (nano-HPLC) system (Dionex), HTC-PAL autosampler (CTC Analytics), and a nanoelectrospray ion source. Five microliters of each sample were injected × ACKNOWLEDGMENTS. This work was supported by funding from the Okinawa into a Zorbax 300SB C18 capillary column (0.3 150 mm; Agilent Technol- Institute of Science and Technology (to T.T.), grants from the Japan Society for ogies) and heated at 40 °C. A 1-h HPLC gradient was employed (1% B to 32% the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (16H04675), B in 45 min, 32% B to 45% B in 15 min, with final wash at 75% B for 5 min a JSPS Core-to-Core Program (A) Advanced Research Networks grant, a research and reequilibration at 1% B for 10 min.) using 0.1% FA in distilled water as grant from the Takeda Foundation (to S.T.), and a grant from the European solvent A, and 0.1% FA in ACN as solvent B. A flow rate of 3.5 μL/min was Research Council (to R.J.). We thank Steven D. Aird for English editing.

1. C. Biesemann et al., Proteomic screening of glutamatergic mouse brain synaptosomes 21. N. J. Pavlos et al., Quantitative analysis of synaptic vesicle Rabs uncovers distinct yet isolated by fluorescence activated sorting. EMBO J. 33, 157–170 (2014). overlapping roles for Rab3a and Rab27b in Ca2+-triggered exocytosis. J. Neurosci. 30, 2. S. Takamori et al., Molecular anatomy of a trafficking organelle. Cell 127,831–846 (2006). 13441–13453 (2010). 3. A. Bayés et al., Comparative study of human and mouse postsynaptic proteomes finds 22. N. J. Grimsey, L. J. Coronel, I. C. Cordova, J. Trejo, Recycling and endosomal sorting of high compositional conservation and abundance differences for key synaptic pro- -activated receptor-1 is distinctly regulated by Rab11A and Rab11B proteins. teins. PLoS One 7, e46683 (2012). J. Biol. Chem. 291, 2223–2236 (2016). 4. J. Boyken et al., Molecular profiling of synaptic vesicle docking sites reveals novel 23. M. Forgac, Vacuolar ATPases: Rotary proton pumps in physiology and pathophysiol- proteins but few differences between glutamatergic and GABAergic synapses. Neu- ogy. Nat. Rev. Mol. Cell Biol. 8, 917–929 (2007). ron 78, 285–297 (2013). 24. M. Merkulova et al., Mapping the H(+) (V)-ATPase interactome: Identification of proteins 5. N. A. O’Rourke, N. C. Weiler, K. D. Micheva, S. J. Smith, Deep molecular diversity of mammalian involved in trafficking, folding, assembly and phosphorylation. Sci. Rep. 5, 14827 (2015). synapses: Why it matters and how to measure it. Nat. Rev. Neurosci. 13, 365–379 (2012). 25. A. César-Razquin et al., A call for systematic research on solute carriers. Cell 162, 6. B. G. Wilhelm et al., Composition of isolated synaptic boutons reveals the amounts of 478–487 (2015). vesicle trafficking proteins. Science 344, 1023–1028 (2014). 26. S. Takamori, J. S. Rhee, C. Rosenmund, R. Jahn, Identification of a vesicular glutamate trans- 7. M. Grønborg et al., Quantitative comparison of glutamatergic and GABAergic syn- porter that defines a glutamatergic phenotype in neurons. Nature 407,189–194 (2000). aptic vesicles unveils selectivity for few proteins including MAL2, a novel synaptic 27.M.J.Nirenberg,Y.Liu,D.Peter,R.H.Edwards,V.M.Pickel,Thevesicularmonoamine vesicle protein. J. Neurosci. 30,2–12 (2010). transporter 2 is present in small synaptic vesicles and preferentially localizes to large dense 8. Y. C. Li, N. L. Chanaday, W. Xu, E. T. Kavalali, Synaptotagmin-1- and synaptotagmin-7- core vesicles in rat solitary tract nuclei. Proc. Natl. Acad. Sci. U.S.A. 92, 8773–8777 (1995). dependent fusion mechanisms target synaptic vesicles to kinetically distinct endocytic 28. S. M. Ferguson et al., Vesicular localization and activity-dependent trafficking of pathways. Neuron 93, 616–631.e3 (2017). presynaptic choline transporters. J. Neurosci. 23, 9697–9709 (2003). 9. S. L. Jackman, J. Turecek, J. E. Belinsky, W. G. Regehr, The calcium sensor synapto- 29. E. Weihe, J. H. Tao-Cheng, M. K. Schäfer, J. D. Erickson, L. E. Eiden, Visualization of the vesicular tagmin 7 is required for synaptic facilitation. Nature 529,88–91 (2016). acetylcholine transporter in cholinergic nerve terminals and its targeting to a specific pop- 10. H. Liu et al., Synaptotagmin 7 functions as a Ca2+-sensor for synaptic vesicle re- ulation of small synaptic vesicles. Proc.Natl.Acad.Sci.U.S.A.93, 3547–3552 (1996). plenishment. eLife 3, e01524 (2014). 30. R. Janz, K. Hofmann, T. C. Südhof, SVOP, an evolutionarily conserved synaptic vesicle protein, 11. R. Aebersold, M. Mann, Mass-spectrometric exploration of proteome structure and suggests novel transport functions of synaptic vesicles. J. Neurosci. 18, 9269–9281 (1998). function. Nature 537, 347–355 (2016). 31. M. Larhammar et al., SLC10A4 is a vesicular amine-associated transporter modulating 12. P. G. Righetti, E. Boschetti, Sherlock Holmes and the proteome–A detective story. FEBS dopamine homeostasis. Biol. Psychiatry 77, 526–536 (2015). J. 274, 897–905 (2007). 32. W. Kostich et al., Inhibition of AAK1 kinase as a novel therapeutic approach to treat 13. A. J. Alpert, Electrostatic repulsion hydrophilic interaction chromatography for iso- neuropathic pain. J. Pharmacol. Exp. Ther. 358, 371–386 (2016). cratic separation of charged solutes and selective isolation of phosphopeptides. Anal. 33. L. Y. Wang, L. K. Kaczmarek, High-frequency firing helps replenish the readily re- Chem. 80,62–76 (2008). leasable pool of synaptic vesicles. Nature 394, 384–388 (1998). 14. A. J. Alpert et al., Peptide orientation affects selectivity in ion-exchange chroma- 34. W. B. Huttner, W. Schiebler, P. Greengard, P. De Camilli, I (protein I), a nerve tography. Anal. Chem. 82, 5253–5259 (2010). terminal-specific phosphoprotein. III. Its association with synaptic vesicles studied in a 15. F. Koopmans et al., SynGO: An evidence-based, expert-curated knowledge base for highly purified synaptic vesicle preparation. J. Cell Biol. 96, 1374–1388 (1983). the synapse. Neuron 103, 217–234.e4 (2019). 35. M. Fromer et al., De novo mutations in schizophrenia implicate synaptic networks. 16. R. Jahn, T. C. Südhof, Synaptic vesicles and exocytosis. Annu. Rev. Neurosci. 17, Nature 506, 179–184 (2014). 219–246 (1994). 36. R. Reig-Viader, C. Sindreu, À. Bayés, Synaptic proteomics as a means to identify the 17. M. Morciano et al., Immunoisolation of two synaptic vesicle pools from synapto- molecular basis of mental illness: Are we getting there? Prog. Neuropsychopharmacol. somes: A proteomics analysis. J. Neurochem. 95, 1732–1745 (2005). Biol. Psychiatry 84,353–361 (2018). 18. N. Hirokawa, K. Sobue, K. Kanda, A. Harada, H. Yorifuji, The cytoskeletal architecture of the 37. P. F. Sullivan, M. J. Daly, M. O’Donovan, Genetic architectures of psychiatric disorders: presynaptic terminal and molecular structure of synapsin 1. J. Cell Biol. 108, 111–126 (1989). The emerging picture and its implications. Nat. Rev. Genet. 13, 537– 551 (2012). 19. R. Jahn, D. Fasshauer, Molecular machines governing exocytosis of synaptic vesicles. 38. A. P. Wingo et al., Large-scale proteomic analysis of identifies proteins Nature 490, 201–207 (2012). associated with cognitive trajectory in advanced age. Nat. Commun. 10, 1619 (2019). 20. H. Stenmark, Rab GTPases as coordinators of vesicle traffic. Nat. Rev. Mol. Cell Biol. 10, 39. R. D. Emes et al., Evolutionary expansion and anatomical specialization of synapse 513–525 (2009). proteome complexity. Nat. Neurosci. 11, 799–806 (2008).

33596 | www.pnas.org/cgi/doi/10.1073/pnas.2011870117 Taoufiq et al. Downloaded by guest on October 1, 2021