Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 61 (5), 734-49

An Unbiased Expression Screen for Synaptogenic Proteins Identifies the LRRTM Protein Family as Synaptic Organizers


An Unbiased Expression Screen for Synaptogenic Proteins Identifies the LRRTM Protein Family as Synaptic Organizers

Michael W Linhoff et al. Neuron.


Delineating the molecular basis of synapse development is crucial for understanding brain function. Cocultures of neurons with transfected fibroblasts have demonstrated the synapse-promoting activity of candidate molecules. Here, we performed an unbiased expression screen for synaptogenic proteins in the coculture assay using custom-made cDNA libraries. Reisolation of NGL-3/LRRC4B and neuroligin-2 accounts for a minority of positive clones, indicating that current understanding of mammalian synaptogenic proteins is incomplete. We identify LRRTM1 as a transmembrane protein that induces presynaptic differentiation in contacting axons. All four LRRTM family members exhibit synaptogenic activity, LRRTMs localize to excitatory synapses, and artificially induced clustering of LRRTMs mediates postsynaptic differentiation. We generate LRRTM1(-/-) mice and reveal altered distribution of the vesicular glutamate transporter VGLUT1, confirming an in vivo synaptic function. These results suggest a prevalence of LRR domain proteins in trans-synaptic signaling and provide a cellular basis for the reported linkage of LRRTM1 to handedness and schizophrenia.


Figure 1
Figure 1. Expression Screening Identifies LRRTM1 as a Synaptogenic Factor
A. Construction of cDNA expression libraries. Flow diagram summarizes the procedure. The autoradiograph in the centre shows the size range of the second-strand cDNA product. Characteristics of the libraries are summarized in the table. B. Flow diagram illustrating the experimental protocol of the expression screen leading to the discovery of LRRTM1. Expression library pools were minimally amplified in soft agar and plasmid DNA prepared and transfected into COS cells in a 12-well tissue culture plate. Transfected COS cells were seeded onto neuron coverslips, and the co-culture coverslips were transferred back into the astrocyte feeder dish. After ∼24 hours, co-culture coverslips were fixed, immunolabeled, and scanned visually on a fluorescent microscope. Images are shown from the original pool, PC064, that scored as positive and was later found to contain LRRTM1. Synaptogenic activity is indicated by the presence of synapsin clusters (red) on axons contacting a COS cell; these presumed induced clusters did not label for postsynaptic markers PSD-95 family plus gephyrin (green), unlike the endogenous neighboring synapses on dendrites. This pool tested negative for PCR tests for known synaptogenic proteins and was further broken down. Sequence analysis revealed the single active clone to be rat LRRTM1. The final cartoon illustrates LRRTM domain structure. LRRTMs contain ten extracellular leucine rich repeats (green) flanked by N- and C-terminal disulfide bonded “capping” domains (yellow), a single pass transmembrane domain (red), and a potential PDZ ligand (magenta).
Figure 2
Figure 2. Quantitation of the Synaptogenic Activity of LRRTM Family Members
A-E. COS cells were transfected with C-terminal CFP fusion proteins, co-cultured with hippocampal neurons, and the clustering of synapsin was quantitated at sites where hippocampal axons contacted transfected cells and not dendrites. A,B. Synapsin clustering is observed along axons (labeled with dephosphorylated tau) contacting COS cells expressing LRRTM1-CFP or LRRTM2-CFP and lacking contact with dendrites (labeled with MAP2). The insets show apposition of LRRTM-CFP clusters in the COS cells and synapsin clusters in contacting axons. C. No significant synapsin clustering is evident along axons contacting AMIGO-CFP expressing COS cells. The only synapsin clusters occur at axon-dendrite contacts, at endogenous synapses. D. Quantitation of the total integrated intensity of synapsin staining associated with COS cells transfected with the indicated CFP fusion proteins and not associated with MAP2. E. Quantitation of the total synapsin pixel area associated with COS cells transfected with the indicated CFP fusion proteins divided by the tau-positive axon contact area. ANOVA p<0.0001 for D,E. Scale bar = 10 μm. F-J. COS cells expressing LRRTM-CFP fusion proteins and co-cultured with hippocampal neurons clustered the specific glutamatergic presynaptic marker VGLUT1 at non-synaptic sites in contacting axons. F. For quantitating VGLUT1 clustering, individual transfected COS cells were scored as either positive or negative; one hundred cells were assayed per coverslip, and four coverslips were analyzed for each recombinant protein. ANOVA p<0.0001. G-J. Expression of LRRTM1-CFP (G), LRRTM2-CFP (H), LRRTM3-CFP (I) or LRRTM4-CFP (J) results in VGLUT1 clustering on axons contacting the expressing COS cell and not associated with PSD-95-family proteins. This is a greater level of clustering than is generally seen for LRRTM3. Scale bar = 10 μm.
Figure 3
Figure 3. LRRTM Instructs Assembly of Functional Vesicle Release Sites
HEK cells were co-transfected with LRRTM2-CFP and the NMDA receptor subunits, YFP-NR1 and NR2A. A similar co-transfection was performed with N-cadherin-CFP and NMDA receptor subunits as a negative control. The transfected HEK cells were co-cultured with hippocampal neurons for ∼24 hours before recording. A. Representative trace from whole cell recording of a HEK cell co-transfected with LRRTM2-CFP and NMDAR. Spontaneous synaptic-like currents can be seen. TTX abolished this bursting activity, leaving miniature EPSC-like events. The inset shows sample miniature EPSC-like events and an average of such events from one cell. B. Trace from whole cell recording of one of the most active HEK cells co-transfected with N-cadherin-CFP and NMDAR. Only occasional isolated events were observed in the absence or presence of TTX; inset shows a sample event in the presence of TTX. C. Frequency plot of EPSC-like events comparing LRRTM2-CFP and N-cadherin-CFP transfected HEK cells in the absence and presence of TTX. Number of cells recorded: LRRTM2 (22), N-cadherin (17), LRRTM2 + TTX (11), N-cadherin + TTX (10). D. Cumulative probability plot of current amplitude for the four conditions shown in panel C.
Figure 4
Figure 4. The LRR Domain of LRRTM Instructs Presynaptic Differentiation
A. Right panel: clustering of synapsin was observed at sites where axons contacted beads coated with LRRTM2 LRR-AP. Left panel: no clustering was observed at sites where axons contacted control beads coated with control AP protein. Top panel: phase contrast, Bottom panel with inset: synapsin immunostaining was used to detect presynaptic differentiation, tau labels axons, and the phase contrast bead image is merged in the blue channel. B. Right panel: clustering of VGLUT1 was observed at sites where axons contacted beads coated with LRRTM2 LRR-AP. Left panel: no clustering was observed at the axon-contact site with control beads. Top panel: phase contrast, Bottom panel: immunostaining for SynGAP was used to detect endogenous synapses, and VGLUT1 immunostaining was used to detect excitatory presynaptic differentiation. In the inset, bassoon and VGLUT1 immunostaining show the separation of the synaptic vesicle pool and the active zone at the site of axon contact with beads (phase contrast image placed in blue channel). Scale bar = 10 μm.
Figure 5
Figure 5. Recombinant LRRTMs Localize to Excitatory Postsynaptic Sites Independently of PDZ Domain Binding
A-C. Hippocampal neurons were transfected at 8-9 DIV with YFP-LRRTM1 (A,B) or YFP-LRRTM2 (C), and the cultures were fixed 2-3 days after transfection. YFP-LRRTMs are localized to puncta in the dendrites of hippocampal neurons. YFP-LRRTM puncta co-localize with puncta of PSD-95 family excitatory postsynaptic proteins but not gephyrin inhibitory postsynaptic protein (A,C). YFP-LRRTMs were not observed in axons or at presynaptic sites; panel B shows YFP-LRRTM1 co-localized with PSD-95 family and apposed to VGLUT1. Scale bar = 10 μm. D-F. Recombinant LRRTM and PSD-95 interact in COS cells. D. Confocal image of a COS cell co-transfected with YFP-LRRTM2 and PSD-95-mRFP. The two proteins form co-localized aggregates. E. Confocal image of a COS cell co-transfected with YFP-LRRTM2ΔPDZ and PSD-95-mRFP. While YFP-LRRTM2ΔPDZ forms small clusters even when expressed alone, PSD-95 does not co-cluster but remains diffusely localized. Scale bar = 10 μm. F. Myc-tagged PSD-95 was co-transfected into COS cells with either wild type YFP-LRRTM2 or YFP-LRRTM2ΔPDZ. Immunoprecipitation with anti-GFP antibody pulled down Myc-PSD-95 only when the full C-terminus of LRRTM2 was present. G-I. Hippocampal neurons were transfected with recombinant YFP-LRRTM2 constructs as in panels A-C. G. Clustering of YFP-LRRTM2 at postsynaptic sites colocalizing with PSD-95 is again shown. H. Deletion of the last 4 residues, the PDZ domain binding site, did not abolish synaptic clustering. I. In contrast, a larger deletion of the C-terminus leaving only 17 residues after the transmembrane domain abolished synaptic clustering. The C-terminal deletion still reached the neuron surface (see Figure S6). Scale bar = 10 μm.
Figure 6
Figure 6. Clustering of LRRTMs Mediates Postsynaptic Differentiation
Neurons were transfected with YFP-LRRTM1 (A) or YFP-LRRTM2 (B-D) and the YFP fusion protein was artificially clustered using beads coated with antibodies to GFP. The 1 μm diameter beads are visible in the phase contrast images. The anti-GFP beads effectively cluster the YFP fusion proteins. A,B. Staining for the NMDA-type glutamate receptor essential subunit NR1 shows clustering of NR1 at bead induced clusters of YFP-LRRTMs, at non-synaptic sites lacking VGLUT1. Additional NR1 clusters are observed at bona fide postsynaptic sites associated with VGLUT1. The boxed regions are shown magnified in the bottom panels. Scale bar = 10 μm. C. PSD-95 family proteins also cluster at bead induced clusters of YFP-LRRTM, at non-synaptic sites lacking VGLUT1. D. The inhibitory postsynaptic marker gephyrin shows no clustering by bead clustered YFP-LRRTM.
Figure 7
Figure 7. Localization of LRRTM2 to Excitatory Synapses In Vivo
A. An antibody against LRRTM2 was generated. The blot shows recognition of an ∼80 kDa protein in brain P2 lysate and in HEK cells transfected for recombinant LRRTM2 but not LRRTM1. B. Synaptic fractionation of adult rat brain homogenate. The ∼80 kDa band corresponding to LRRTM2 is abundant in the PSD fraction along with PSD-95. LRRTM2 is detected in the P2, Synaptosome, and Synaptic Plasma Membrane (SPM) fractions but less so in the Cytoplasm and Synaptic Vesicle (SV) fractions. C-F. Confocal analysis of LRRTM2 immunofluorescence in mouse brain sections. C. LRRTM2 is widely detected in synaptic neuropil regions throughout the cortex and hippocampal formation. LRRTM2 exhibits a laminar-selective distribution; for example, in the CA1 region, LRRTM2 is present at higher levels in stratum lacunosum moleculare (SLM) than in radiatum (Rad) or oriens (Ori). D. LRRTM2 is detected in the granule cell layer and molecular layer of the cerebellum. C,D scale bar = 200 μm. E. In hippocampus, LRRTM2 is particularly prominent in CA3 mossy fibers synapses shown here. Co-localization with the general synaptic marker bassoon and the excitatory synaptic marker VGLUT1 is shown. Enlarged regions are shown in the same color channels as the main panels (upper left LRRTM2 greyscale; lower left LRRTM2 green with bassoon blue; upper right LRRTM2 green with VGLUT1 red; lower right triple channel overlay). F. LRRTM2 also co-localized with bassoon and VGLUT1 in cerebellar glomerular synapses. E,F Scale bar = 20 μm main panels, 5 μm enlarged regions.
Figure 8
Figure 8. Normal Brain Morphology but Altered VGLUT1 Immunofluorescence in LRRTM1 -/- Mice
A. Targeting strategy for Lrrtm1 locus. Part of the wild-type Lrrtm1 locus as well as the elements of the replacing cassette are shown. The locations of the restriction enzyme cleavage sites used in the Southern blotting analysis are indicated: B, BamHI; E, EcoRI; IRES, internal ribosome entry site. B. Southern blotting results with a correctly targeted ES cell clone (+/-) and with a wild-type (+/+) ES cell clone are shown. Fragment lengths expected are 6 kb for wild type and 3.6 kb for a correctly targeted allele with the 5′ probe (EcoRI digest), and 10.5 kb for wild type and 5.4 kb for a correctly targeted allele with the 3′ probe (BamHI and SpeI digest). C. Genotyping results for wild-type (+/+) mouse and mice heterozygous (+/-) or homozygous (-/-) for Lrrtm1 deletion. D. No differences were detected in overall cellular or synaptic distribution in the LRRTM1 -/- mouse brain compared with wild type. DAPI stain for nuclei shows normal cellular organization in the hippocampal formation. Immunofluorescence patterns for the presynaptic active zone protein bassoon are indistinguishable between LRRTM -/- and wild type hippocampus. Scale bar = 200 μm. E-G. Quantitative analysis of immunofluorescence in the hippocampal formation at 4-5 weeks. E. Sample confocal images, here from CA1 stratum radiatum. Scale bar = 10 μm. F. There was no difference in bassoon immunopositive puncta area between LRRTM1 -/- and wild type in any region measured. Measures were made from the same multi-channel images as for panel G. G. Puncta area for the vesicle-associated protein VGLUT1 was selectively increased in CA1 stratum radiatum (Rad) and oriens (Ori) (t-test p<0.01, duplicate experiments of 2 LRRTM1 -/- and 3 wild type mice), but unchanged in LRRTM1 -/- relative to wild type mice in CA1 stratum lacunosum moleculare (SLM) and CA3 stratum lucidum (Luc). Mean values for VGLUT1 puncta area in wild type mice were 1.01 μm2 CA3 Luc, 0.50 μm2 CA1 SLM, 0.49 μm2 CA1 Rad, 0.50 μm2 CA1 Ori. Panels A-C are reproduced from a doctoral dissertation (Lauren, 2007).

Comment in

Similar articles

See all similar articles

Cited by 157 articles

See all "Cited by" articles

Publication types

MeSH terms