Ferric Chelate Reductase 1 Like Protein (FRRS1L) Associates with Dynein Vesicles and Regulates Glutamatergic Synaptic Transmission

In the brain, AMPA receptors (AMPARs)-mediated excitatory synaptic transmission is critically regulated by the receptor auxiliary subunits. Recent proteomic studies have identified that Ferric Chelate Reductase 1 Like protein (FRRS1L), whose mutations in human lead to epilepsy, choreoathetosis, and cognitive deficits, is present in native AMPAR complexes in the brain. Here we have characterized FRRS1L in both heterologous cells and in mouse neurons. We found that FRRS1L interacts with both GluA1 and GluA2 subunits of AMPARs, but does not form dimers/oligomers, in HEK cells. In mouse hippocampal neurons, recombinant FRRS1L at the neuronal surface partially co-localizes with GluA1 and primarily localizes at non-synaptic membranes. In addition, native FRRS1L in hippocampus is localized at dynein, but not kinesin5B, vesicles. Functionally, over-expression of FRRS1L in hippocampal neurons does not change glutamatergic synaptic transmission. In contrast, single-cell knockout (KO) of FRRS1L strongly reduces the expression levels of the GluA1 subunit at the neuronal surface, and significantly decreases AMPAR-mediated synaptic transmission in mouse hippocampal pyramidal neurons. Taken together, these data characterize FRRS1L in heterologous cells and neurons, and reveal an important role of FRRS1L in the regulation of excitatory synaptic strength.

Recent proteomic studies have identified several uncharacterized proteins, including transmembrane, GPIanchored and secreted molecules, that are associated with endogenous AMPARs in the brain (von Engelhardt et al., 2010;Schwenk et al., 2012;Chen et al., 2014). Understanding the function of these molecules in the regulation of AMPARs will be important for elucidation of the regulatory mechanisms for controlling excitatory synaptic transmission. One of these proteins is FRRS1L, which is also named C9orf4 (Schwenk et al., 2012). Human and mouse genetic studies have shown a critical role of FRRS1L in human and animal behavior (Madeo et al., 2016;Shaheen et al., 2016). Indeed, loss-of-function mutations of FRRS1L in human lead to epilepsy, prominent choreoathetosis, and severe impairment of cognitive functions (Madeo et al., 2016;Shaheen et al., 2016). Thus, it is important to understand the role of FRRS1L in the regulation of synaptic transmission.
Here we have performed basic characterizations of FRRS1L in heterologous cells and in neurons, and have employed gain-offunction and loss-of-function approaches to examine the role of FRRS1L in the regulation of glutamatergic synaptic transmission. We find that over-expressed FRRS1L in hippocampal neurons only co-localizes with a portion of the AMPAR GluA1 subunit at the plasma membrane, suggesting that FRRS1L might be involved in the regulation of a subpopulation of AMPARs in neurons. Interestingly, native FRRS1L is localized at dynein, but not kinesin KIF5B, vesicles. Functionally, FRRS1L does not regulate AMPAR abundance at the cell surface in HEK cells, and similarly, over-expression of FRRS1L in culture hippocampal neurons does not change the strength of glutamatergic synaptic transmission. In contrast, single guide RNA (sgRNA)-based single-cell knockout (KO) of FRRS1L in hippocampal neurons leads to a modest, but significant, reduction of AMPAR-mediated synaptic transmission. Importantly, the sgRNA-mediated effect can be fully rescued by a sgRNA resistant FRRS1L construct in neurons. Taken together, these data reveal a distinct subcellular distribution of FRRS1L in hippocampal neurons and demonstrate an important role of FRRS1L in the regulation of AMPAR-mediated synaptic transmission in hippocampus.

Mouse Genetics
This study was carried out in accordance with the recommendations of the Animal Care and Use Committee (ACUC) guidelines at National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health (NIH). The animal protocols were approved by the NINDS ACUC at NIH (protocol number: 1339-15). Mice of either sex were used. Animals were given access to food and water ad libitum.

Mouse Tissue Preparations and Western Blotting
Mouse tissues of interest (6-8-week old for tissue distribution experiments) were quickly dissected after anesthetizing the donor mice with isoflurane (Baxter, NDC10019-360-40), then placed in 2.0 ml microfuge tubes and kept in dry ice to snap freeze. The samples were separately added into an ice-cold glass pestle containing 1 ml ice-cold non-detergent homogenization buffer [250 mM sucrose, 1 mM EDTA, 10 mM Tris-HCl buffer (pH 7.2), and EDTA-free protease inhibitors (1 tablet/10 ml, Roche, 5892791001)] to homogenize. The homogenate was then sonicated with two 10-s pulses (30 s pause in between pulses) using an ice-cold ultrasonic probe, and then centrifuged at 1,000 × g for 10 min to remove neuronal nuclei and other cell debris. The crude post-nuclear supernatant fraction (PNS) was transferred to a 1.5 ml ultracentrifuge tube; the pellet containing the nuclei was discarded. The PNS was centrifuged at 100,000 × g for 2 h at 4 • C to collect membrane fractions in the pellet. The membrane fractions were lysed in appropriate volume of ice-cold lysis buffer containing 25 mM Tris (pH 7.4), 1% Triton X-100, 150 mM NaCl, 5% glycerol, 1 mM EDTA, and EDTA-free protease inhibitors. Equal amounts of lysates measured by BCA protein assay kit (Thermofisher, Cat. 23225) were subjected to SDS-PAGE and immunoblotting with an anti-FRRS1L (1:1,000, Santa Cruz, sc-398692) or anti-α-tubulin antibody (1:5000, Sigma, T8203). All data were collected from at least three independent experiments.

Immunocytochemistry in HEK Cells
For surface and total AMPAR subunit expression, HEK cells (2 × 10 4 cells/coverslip) cultured on coverslips were transfected with Flag-GluA1 or Flag-GluA2 on their own or together with pCAGGS/FRRS1L-IRES-mCherry by effectene transfection reagent, and incubated in 37 • C incubator containing 5% CO 2 for approximate 40-48 h. After removing the cell media, cells were washed once by 1xPBS and then fixed with ice-cold fixation buffer (10 ml 16% paraformaldehyde solution, 8 ml 5× PBS and 1.6 g sucrose dissolved in 22 ml distilled water) for 15 min. Cells were then blocked by PBS containing 10% normal goat serum (NGS) (Vector Laboratories) for 30 min. Without permeabilization, cells were incubated with a monoclonal anti-Flag antibody (1:1,000) dissolved in 1× PBS containing 3% NGS at room temperature (RT) for 2 h for surface Flag (sFlag) tag labeling. After three times washing by 1x PBS, cells were permeabilized with 0.2% Triton X-100 for 15 min, and then blocked for 30 min in 1× PBS containing 10% NGS. Cells were then incubated with a polyclonal anti-Flag antibody (1:1,000) at 4 • C overnight for total Flag (tFlag) tag labeling. Alexa Fluor (AF) 647 anti-mouse and AF488 anti-rabbit secondary antibodies (Molecular Probe) were used to label sFlag and tFlag. For co-localization of surface and total HA-FRRS1L or sFlag-GluA1/GluA2 in HEK cells, the cells were transfected and treated as above with indicated plasmids, and were subjected to similar immunolabeling procedures as described above. Coverslips were mounted with Fluoromount G (Southern Biotech) for imaging acquisition.

Image Acquisition
For immunostaining in HEK cells, images were acquired on a Zeiss LSM 880 laser scanning confocal microscope using a 63x oil objective (1.4 numerical aperture). Multiple z sections (9 optical slices) were acquired at 0.5-1.0 µm intervals. Images were captured using a 1024 × 1024 pixel screen, and gains for both fluorophores were set between 700 and 800. For immunostaining in neuronal cultures, images were acquired on a Zeiss LSM 880 laser scanning confocal microscope using a 63x oil objective (1.4 numerical aperture). Image acquisition was performed with identical settings for a specific experiment. Multiple z sections of secondary apical dendrites were collected at 1.0-1.5 µm intervals with 512 × 512 pixel screen. Pinhole was set to 1 airy unit for all experiments. Scan speed function was set to 9 and the mean of 4 lines was applied. Laser power, digital gain, and offset settings were all identical in each experiment by using the "reuse" function in LSM software.

Images Analysis
For quantitative analysis of immunostaining in HEK cells and in neurons, maximal projection images were created with ZEN software (Zeiss) from 4 to 6 serial optical sections. For each image collected containing sGluA1 or PSD95, dendritic outline was drawn to cover 20-30 µm in length (representing a surface area of 850-1,000 pixels). The integrated fluorescent intensity targeting protein was calculated from one segment of dendrite positive for HA-FRRS1L. Using Metamorph (Universal Imaging Corp.), quantitation of the fluorescence signal from sFlag or tFlag in HEK cells and surface/total AMPAR subunits in neurons were determined from fluorescent signal above a threshold. The threshold value was held identical within single experiments, and only slightly adjusted between independent experiments. For surface/total ratio of GluA1 staining and GFP fluorescent signals in dissociated hippocampal neuronal cultures, maximal projection images were generated by the LSM880 Brower software. Background was subtracted by using the "subtract background" function in ImageJ software (NIH), and the background level was held identical for all cells within each experiment. Region of interest (ROI) was defined along a segment of the dendrite 20-30 µm according to the fluorescence signal distinguished from the background in ImageJ software. Average values of fluorescence intensities in ROI (the total fluorescent intensity divided by the total area of a dendritic segment) were calculated by ImageJ.
For co-localization analysis in cultured hippocampal neurons, 25-30 µm secondary dendritic segments were chosen as ROI. HA-FRRS1L, GluA1 or PSD-95 puncta were thresholded and confirmed visually to select appropriate clusters following a minimal size cut-off, which included all recognizable clusters. The co-localization percentage was quantified by the measurement of HA-FRRS1L-positive GluA1 or PSD-95 puncta divided by total number of thresholded GluA1 or PSD-95 puncta. 10-27 dendrites were analyzed from approximately 5-10 neurons.

In Situ Hybridization (ISH) in Cultured Neurons
Cultured neurons on coverslips were transfected with sgRNAs at DIV2 and fixed in fixation buffer at RT for 15 min at DIV16. Coverslips were then washed three times with 1× PBS and sequentially incubated in 50, 70, and 100% EtOH for 5 min for dehydration. The coverslips were then submerged in 70 and 50% EtOH for 2 min for rehydration followed by keeping in 1× PBS for 10 min. The coverslips were then treated with hydrogen peroxide and protease III (1:15 diluted in 1× PBS) (RNAscope, 2000258) for 10 min followed by washing twice in 1× PBS. The coverslips were quickly transferred to the barriered SuperFrost Plus slide. All of the above steps were operated at RT. Four drops (∼50-80 µl) of FRRS1L probe (RNAscope, 16137A) were added to submerge the coverslip, then the slide was incubated in a humidity control tray at 40 • C for 2 h followed by washing twice in 1× wash buffer (RNAscope, 2000003) for 2 min at RT. The slide was then respectively incubated with detection reagents AMP1-6 and Fast RED reagent (A:B, 1:60 ratio) (RNAscope, 2000173) as described in RNAscope HD 2.5 RED reagent protocol (322360-USM). After rinsing in 1x wash buffer, the slide was removed from the humidity control tray and dried in dry oven at 60 • C for 20 min. The slide containing coverslip was then dipped in fresh Xylene and 3 drops of DAPI permount media were quickly added followed by covering with a 24 mm × 50 mm coverslip without air bubbles.
Positive punctate signal was detected under a standard bright field microscope at 63x magnification using confocal microscope. GFP positive signals and red punctate dots were presented in FITC and 550 nm channels. The staining of ISH was categorized into five grades based on the recommendation from the RNAscope manufacturer (322360-USM): 0, 1+, 2+, 3+, and 4+. Staining Scores were defined by the following criteria: 0, no staining or less than 1 dot per 10 cells; 1, 1-3 dots per cell; 2, 4-10 dots per cell, very few dot clusters; 3, >10 dots per cell, less than 10% positive cells have dot clusters; 4, >10 dots per cell, more than 10% positive cells have dot clusters.

In Utero Electroporation
Timed-pregnant WT C57B6 mice at embryonic 14.5 days (E14.5) were anesthetized with isoflurane. The abdominal cavity was opened and 7-9 embryos in the uterine horns were gently exposed. The lateral ventricle of each embryo was manually injected with approximately 1-2 µl of FRRS1L sgRNA plasmid at a concentration of 2 µg/µl mixed with 0.05% fast green. The pipettes were beveled with a BV-10 micropipette Beveller (Sutter) before microinjection. After unilateral injection in embryos, voltage steps via tweezertrodes (5 mm round, platinum electrodes and BTX electroporator, BTX, ECM830) positioned on either side of the embryo head were applied across the uterus to target hippocampal neural progenitors. Voltage was set at 45 V for 5 pulses at 1 Hz, each pulse lasting 50 ms. The embryos were then moistened with sterilized PBS (pre-warmed at 37 • C) and returned to the abdominal cavity. Buprenex (0.1 mg/kg) was then applied and the wound was subsequently sutured.

Electrophysiology in Hippocampal Slices
Organotypic hippocampal slice cultures were prepared and transfected as previously described (Gu et al., 2016a). Both sexes of C57B6 mice at the age of p6-p8 were used for organotypic hippocampal slice cultures. About 2-4 days after culture, slices were transfected with FRRS1L-IRES-EGFP or with FRRS1L sgRNA #1 or sgRNA #3 plasmids by gene gun-mediated transfection. Slices were cultured for an additional 2-5 days for over-expression or 14-16 days for sgRNA transfection. After transfection, slices were transferred to a submersion chamber on an upright Olympus microscope, perfused in normal ACSF containing (in mM) NaCl 119, KCl 2.5, NaHCO 3 26.2, NaH 2 PO 4 1, glucose 11, CaCl 2 4.0, and MgSO 4 4.0, with picrotoxin (100 µM), and saturated with 95% O 2 /5% CO 2 . For recording evoked EPSCs in organotypic slices, ACSF was also supplemented with 5-20 µM 2-chloroadenosine to dampen epileptiform activity. EGFP-positive neurons in organotypic slice cultures were identified by epifluorescence microscopy. For acute hippocampal slices, mouse pups at p14-p16 after embryonic IUE were euthanized by decapitation. The brain was immediately placed in ice-cold cutting solution containing (in mM) KCl 2.5, CaCl 2 0.5, MgCl 2 7, NaH 2 PO 4 1.25, NaHCO 3 25, glucose 7, ascorbic acid 1.3 and sucrose 210 as previously described (Gu et al., 2016b). The hippocampi were quickly dissected out on an ice-cold platform and immediately placed onto an ice-cold Argrose gel block (5% Argrose). The gel block was quickly glued on the cutting platform containing ice-cold cutting solution saturated with carbogen. 300 µm transverse slices were cut and recovered at 32 • C for 30-60 min. Slices were then maintained in ACSF (modified to containing 2.5 mM CaCl 2 and 1.3 mM MgCl 2 ) at RT for 30-60 min prior to recording. The slice was then transferred to a recording chamber that is mounted on an upright Olympus microscope. The recording chamber was continuously perfused with ACSF containing picrotoxin (100 µM) and saturated with 95% O 2 /5% CO 2 .
All paired recordings involved simultaneous whole cell recordings from one EGFP-positive neuron and a neighboring EGFP-negative control neuron. Cells were recorded with 3to 5-MO borosilicate glass pipettes. The internal solution contained CsMeSO 4 135 mM, NaCl 8 mM, HEPEs 10 mM, Na 3 GTP 0.3 mM, MgATP 4 mM, EGTA 0.3 mM, QX-314 5 mM and spermine 0.1 mM. The stimulus was adjusted to evoke a measurable, monosynaptic EPSC in both cells. AMPA EPSCs were measured at a holding potential of −70 mV, and NMDA EPSCs were measured at +40 mV and at 100 ms after the stimulus, at which point the AMPA EPSC has completely decayed. Paired-pulse ratios (PPRs) were measured by giving two pulses at a 50-ms interval and taking the ratio of the two peaks of the EPSCs from an average of 30-50 sweeps. Series resistance was monitored and not compensated, and cells in which series resistance varied by 25% during a recording session were discarded. Synaptic responses were collected with a Multiclamp 700B amplifier (Axon Instruments, Foster City, CA, United States), filtered at 2 kHz and digitized at 10 kHz. All pharmacological reagents were purchased from Abcam, and other chemicals were purchased from Sigma.
In the scatter plots for simultaneous dual recordings (Figures 4, 6), each open circle represents one paired recording, and the closed circle represents the average of all paired recordings. In the scatter plot, the x-axis represents the EPSC recorded in the control cell, and the y-axis represents the EPSC recorded in the transfected cell. Virtual 1:1 diagonal line is also shown. If the data point falls above the diagonal line, it indicates that the EPSC is higher in the transfected cell. If it falls below the diagonal line, it indicates that the EPSC is higher in the control cell.
For decay time constants of AMPA EPSCs, a single weighted decay time constant was calculated from the area under the peaknormalized current (Cathala et al., 2005), according to t decay = 1 l peak where t 0 was 80 ms after the peak.

Glutamate Puffing Experiments in HEK Cells
For glutamate-evoked whole-cell currents in HEK cells, cells (2 × 10 4 cells/coverslip) were cultured and transfected with GluA1, GluA2, GluA1 plus FRRS1L-IRES-GFP (ratio at 1:1), or GluA2 plus FRRS1L-IRES-GFP (ratio at 1:1) (an empty plasmid was used to make the total cDNA amount, 0.4 µg, per transfection). On the day of recording, coverslips with transfected HEK cells were transferred to a submersion chamber on an upright Olympus microscope perfused with external solution containing (in mM) NaCl 140, KCl 5, MgCl 2 1.4, EGTA 5, HEPES 10, NaH 2 PO 4 1, D-glucose 10, and NBQX 0.01, with pH adjusted to 7.4. The internal solution contained CsMeSO 4 135 mM, NaCl 8 mM, HEPEs 10 mM, Na 3 GTP 0.3 mM, MgATP 4 mM, EGTA 0.3 mM, QX-314 5 mM, and spermine 0.1 mM. Glutamate induced whole-cell currents in HEK cells were recorded at −70 mV by local fast application (0.5 s) of 1 mM glutamate and 100 µM cyclothiazide, dissolved in the external solution. The tip of Mini-manifold was placed at 100 µm away from the recorded HEK cells. Whole-cell currents were collected with a Multiclamp 700B amplifier (Axon Instruments, Foster City, CA, United States), filtered at 2 kHz and digitized at 10 kHz. All pharmacological reagents were purchased from Abcam, and other chemicals were purchased from Sigma.

Statistical Analysis
Statistical analysis was performed using GraphPad Prism 7. Comparisons between two groups were performed with twotailed, unpaired t-tests except for paired recording whereby the statistic comparisons were performed with two-tailed, paired t-test. Multiple comparisons with three or more groups were performed with one-way ANOVAs. The D'Agostino and Pearson normality test and the Shapiro-Wilk normality test were performed to test the normality of unpaired data. All of the data passed the normality test (α = 0.05) and came from a Gaussian distribution (p > 0.05). Significance was considered when the p-value < 0.05, 0.01, 0.001 or 0.0001 (indicated as * , * * , * * * , or * * * * , respectively). p-values ≥ 0.05 were considered not significant. The data were presented as Mean ± SEM.

Basic Characterizations of FRRS1L in Mouse
Recent proteomic studies have identified FRRS1L as a component of native AMPAR complexes in the brain (Schwenk et al., 2012;Chen et al., 2014). To confirm FRRS1L-AMPAR association, we performed a co-immunoprecipitation (Co-IP) assay in human embryonic kidney 293T (HEK) cells. We found that HA-tagged FRRS1L (HA-FRRS1L) could be co-immunoprecipitated with Flag-tagged GluA1 (Flag-GluA1) or Flag-tagged GluA2 (Flag-GluA2) subunit of the AMPAR from cells co-transfected with both constructs, but not from control cells transfected with either one (Figure 1A), indicating that FRRS1L can associate with both GluA1 and GluA2 in HEK cells. In addition, we found that in heterologous cells, FRRS1L did not form dimers or oligomers, as Myc-tagged FRRS1L (FRRS1L-Myc) was not coimmunoprecipitated by HA-FRRS1L in HEK cells (Figure 1B).
In mouse, FRRS1L protein was detected in the brain as well as in a few other tissues ( Figure 1C). In the brain, FRRS1L had a broad distribution with high expression observed in hippocampus, cortex and thalamus ( Figure 1D). In addition, expression of FRRS1L in the brain steadily increased during the development with high expression levels reached at ∼postnatal two-to three-week-old ( Figure 1E). In dissociated hippocampal primary neuron cultures, the majority of neuronal surface HA-FRRS1L (sHA-FRRS1L) did not co-localize with an excitatory synapse marker, PSD-95 (Figure 1F), suggesting that FRRS1L may mainly exert its function outside of synapses. In addition, a substantial fraction of sHA-FRRS1L co-localized with a portion of endogenous surface GluA1 (sGluA1) (Figure 1F), indicating that FRRS1L associates with a subpopulation of GluA1-containing AMPARs in hippocampal neurons.

FRRS1L Is Localized at Dynein-Positive Vesicles
Microtubule-based transport by dynein and kinesin motors has been shown to play an important role in the regulation of AMPAR trafficking (Kim and Lisman, 2001;Setou et al., 2002;Kapitein et al., 2010;Hoerndli et al., 2013Hoerndli et al., , 2015Heisler et al., 2014). We were wondering whether FRRS1L shows any specificity to dynein or kinesin vesicles. To this end, we performed immunoisolation of vesicles in floated membrane fractions isolated from detergentfree hippocampal homogenate with sucrose step-gradients and utilized Western blotting to analyze the proteins associated biochemically with the vesicles (Figure 2A). We found that AMPARs, FRRS1L, TARP γ8 and CNIH2 were detected in the PNSM, but not in the 35/40 fraction, suggesting that AMPARs and auxiliary subunits are not associated with the specific 35/40 membrane fraction ( Figure 2B). Interestingly, in the vesicleenriched 8/35 fraction (Encalada et al., 2011), while AMPARs, FRRS1L, and TARP γ8 were detected, CNIH2 was not detectable, indicating a diverse distribution of AMPAR auxiliary subunits in distinct membranous entities ( Figure 2B).
In the 8/35 membrane fraction, an antibody against dynein intermediate chain (DIC) pulled down dynein, but not one of the major kinesin heavy chains in neurons, KIF5B ( Figure 2C). Significantly, we found AMPARs and FRRS1L, but not TARP γ8 and CNIH2, in dynein vesicles, confirming an association of endogenous FRRS1L with dynein vesicles (Figure 2C). Interestingly, an antibody against KIF5B pulled down KIF5B and AMPARs, but none of three auxiliary subunits of AMPARs: FRRS1L, TARP γ8, and CNIH2 ( Figure 2D), suggesting that AMPARs in KIF5B vesicles unlikely contain FRRS1L, TARP γ8 or CNIH2. However, we could not exclude the possibility that the levels of these auxiliary subunits in KIF5B immunoprecipitates in the 8/35 membrane fraction may be below the detection thresholds of corresponding antibodies. Taken together, these biochemical data demonstrate that a portion of native FRRS1L associates with dynein vesicles. These data also indicate the existence of distinct subpopulations of hippocampal AMPARs that are decorated with unique combinations of auxiliary subunits.

FRRS1L Does Not Change the Expression of AMPARs at the Plasma Membrane in HEK Cells
We then sought to determine the role of FRRS1L in the regulation of AMPARs. Membrane proteins that interact with AMPARs often regulate AMPAR trafficking (Jackson and Nicoll, 2011;  Straub and Tomita, 2012). To study FRRS1L-mediated effect on AMPAR trafficking, we first expressed AMPAR subunits on their own or together with FRRS1L in HEK cells and employed an immunocytochemical assay to examine the ratio of surface to total expression levels of AMPAR subunits. In HEK cells, HA-FRRS1L on its own could traffic to the cell surface ( Figure 3A). In addition, sHA-FRRS1L co-localized with surface Flag-GluA1 (sFlag-GluA1) or surface Flag-GluA2 (Q) (sFlag-GluA2) in HEK cells (Figure 3B), indicating that FRRS1L is associated with AMPARs on the cell surface. Furthermore, we found that levels of sFlag-GluA1 or sFlag-GluA2 in HEK cells were not altered by co-expression with FRRS1L-IRES-mCherry (Figures 3C,D). We also co-expressed FRRS1L-IRES-mCherry together with GluA1-IRES-EGFP or GluA2-IRES-EGFP in HEK cells and conducted patch clamp recordings to measure AMPARmediated whole-cell currents. We found that the AMPAR subunit expressed on its own generated the same amount of wholecell currents evoked by 1 mM glutamate (in the presence of cyclothiazide to block receptor desensitization) as the AMPAR subunit and FRRS1L expressed together (Figures 3E,F). Taken together, these data indicate that the abundance of surface AMPAR subunits is not changed by FRRS1L co-expression in HEK cells.

Over-Expression of FRRS1L Does Not Affect AMPAR-Mediated Synaptic Transmission in Hippocampal CA1 Neurons
To study the role of FRRS1L in neurons, we first performed an immunocytochemical assay to measure surface and total levels of GluA1, a major AMPAR subunit expressed in hippocampal neurons (Lu et al., 2009), in dissociated primary hippocampal cultures that were transfected with FRRS1L-IRES-EGFP. Double immunolabeling of surface and total GluA1 demonstrated that over-expression of FRRS1L did not change surface levels, total levels, or surface to total ratio of GluA1 expression (Figure 4A), suggesting that increased expression of FRRS1L is not sufficient to alter AMPAR trafficking to the neuronal surface.
To corroborate the immunocytochemical data, we performed electrophysiological recordings to measure AMPAR-mediated synaptic transmission in neurons over-expressing FRRS1L. Toward this end, we biolistically transfected cultured organotypic hippocampal slices with gold particles that were coated with plasmids expressing FRRS1L-IRES-EGFP. Two to five days after transfection, we performed simultaneous dual whole-cell voltage clamp recordings (dual recordings) that were made from a transfected pyramidal cell (identified by EGFP) and a neighboring EGFP-negative neuron to detect AMPAR-and NMDA receptor (NMDAR)-mediated EPSCs. A stimulating electrode was placed at the Schaffer collateral pathway to evoke EPSCs in both cells. We found that over-expression of FRRS1L did not change either AMPA or NMDA EPSCs (Figures 4B,C), and the weighted decay time constant of the AMPA EPSC ( Figure 4D). In addition, the PPR, a measure of presynaptic neurotransmitter release probability, of AMPA EPSCs in the postsynaptic neurons over-expressing FRRS1L was not changed (Figure 4E). These data indicate that over-expression of FRRS1L in hippocampal pyramidal neurons does not affect the strength of excitatory synaptic transmission.
(Commercially available FRRS1L antibodies were not suitable for immunolabeling of endogenous FRRS1L in neurons, and thus we could not perform immunocytochemical experiments in neurons). We found that in neurons expressing sgRNA #3 , FRRS1L mRNAs were essentially not detectable as compared to FRRS1L mRNAs in non-transfected neurons (Control) or sgRNA #1 -expressing neurons (Figure 5B), indicating that we had successfully achieved single-cell KO of FRRS1L in neurons expressing sgRNA #3 .
To study the role of FRRS1L in AMPAR trafficking to the plasma membrane in neurons, we transfected sgRNA #3 , or sgRNA #1 as a control, at DIV2 (days in vitro) neuronal cultures and measured surface and total GluA1 expression levels at DIV16. We found that surface levels of GluA1 were significantly reduced in neurons expressing sgRNA #3 (Figure 5C). Interestingly, total GluA1 expression levels were also reduced in sgRNA #3 -expressing neurons (Figure 5C), suggesting that FRRS1L is critical for the stability of GluA1 protein in neurons. Furthermore, the surface to total ratio of GluA1 expression was significantly decreased in neurons expressing sgRNA #3 (Figure 5C), indicating that targeting of GluA1 to the neuronal surface is impaired in neurons lacking FRRS1L. In contrast, transfection of sgRNA #1 into neurons changed neither surface nor total GluA1 (Figure 5C). Taken together, these data demonstrate that FRRS1L is critical for GluA1 stability in neurons and also plays a role in GluA1 trafficking to the neuronal surface.

FRRS1L Is Important for AMPAR-Mediated Synaptic Transmission in Vitro and in Vivo
Finally, we examined the effect of single-cell KO of FRRS1L on AMPAR-mediated synaptic transmission. We first biolistically transfected cultured organotypic hippocampal slices with sgRNA #3 or sgRNA #1 plasmids. Approximately 2 weeks after transfection, we performed dual recordings to measure glutamatergic synaptic transmission in pyramidal neurons. Expression of sgRNA #3 specifically reduced the amplitude of AMPA EPSCs by ∼30%, without affecting NMDA EPSC amplitudes ( Figure 6A). In addition, there was no change of PPR (Figure 6C), suggesting that loss of FRRS1L in the postsynaptic neurons does not alter presynaptic neurotransmitter release probability. In contrast, expression of sgRNA #1 did not affect AMPA and NMDA EPSCs nor PPR (Figures 6D,F). Importantly, an sgRNA #3 -resistant FRRS1L mutant (FRRS1L * ) fully rescued the deficit of AMPA EPSCs in pyramidal neurons expressing sgRNA #3 (Figures 6G,H), demonstrating that the effect of sgRNA #3 on AMPA EPSCs is due to the loss of FRRS1L protein. Furthermore, none of the manipulations changed the weighted decay time constants of AMPA EPSCs (Figures 6B,E,I). Taken together, these results show that FRRS1L is critical for AMPARmediated synaptic transmission in hippocampal pyramidal neurons.
To study the role of FRRS1L in the regulation of glutamatergic synaptic transmission in vivo, we electroporated plasmids in utero to sparsely express sgRNA #3 in hippocampal progenitor cells in E14.5 mice embryos to inactivate Frrs1l alleles and then performed dual recordings to measure glutamatergic synaptic transmission in acute hippocampal slices prepared from 2-to 3-week-old mice. We found that single-cell KO of FRRS1L with sgRNA #3 in CA1 pyramidal neurons in acute hippocampal slices significantly decreased AMPA, but not NMDA, EPSCs ( Figure 6J). In addition, there was no change of the weighted decay time constant of AMPA EPSCs and PPR (Figures 6K,L). Thus, FRRS1L is important for AMPAR-mediated synaptic transmission in vivo.
Immunocytochemical data in neuronal cultures show that only a minor fraction of immunolabeling of expressed FRRS1L co-localizes with the excitatory synaptic marker, PSD-95, indicating that FRRS1L is largely localized at non-synaptic subcellular regions in neuronal dendrites. Moreover, there is a substantial co-localization between expressed FRRS1L and endogenous AMPAR subunit GluA1, suggesting that FRRS1L associates with a subpopulation of AMPARs. Interestingly, our data demonstrate that FRRS1L, but not TARP γ8 or CNIH2, associates with dynein, but not kinesin KIF5B, vesicles. Previous studies have shown that microtubule-based motors, kinesin and dynein, are important for the long-range dendritic transport of vesicles containing the assembled AMPARs out of the ER and other compartments of the endomembrane system, and for the regulation of excitatory synaptic transmission (Kim and Lisman, 2001;Setou et al., 2002;Kapitein et al., 2010;Hoerndli et al., 2013). The association of FRRS1L with dynein vesicles suggests that FRRS1L-containing AMPARs could be transported by dynein-dependent mechanisms. However, the exact role of FRRS1L in dynein-based AMPAR trafficking remains to be examined.
of AMPAR subunits together with FRRS1L does not alter 1 mM glutamate-evoked whole-cell currents in the presence of cyclothiazide in HEK cells. Thus, FRRS1L on its own appears not be sufficient to change the expression of AMPARs at the plasma membrane, which is different from other AMPAR auxiliary subunits that often either promote or diminish AMPARs at the cell surface in heterologous cells (Tomita et al., 2004;Schwenk et al., 2009;von Engelhardt et al., 2010;Shanks et al., 2012;Gratacos-Batlle et al., 2014;Gu et al., 2016a;Erlenhardt et al., 2016;Wei et al., 2016). Furthermore, in neurons, over-expression of FRRS1L changes neither the surface GluA1 nor AMPAR-mediated synaptic transmission. These data indicate that increased expression of FRRS1L in neurons is not sufficient to alter AMPAR trafficking to the plasma membrane and FRRS1L is unlikely a limiting factor in determining the expression levels of AMPARs at the surface of cultured neurons in our experiments. Our findings that over-expression of FRRS1L does not change excitatory synaptic transmission and that single-cell genetic deletion of FRRS1L decreases AMPAR-mediated EPSCs resemble previous work on CNIH2. In hippocampal CA1 neurons, it has been reported that over-expression of CNIH2 in hippocampal neurons does not change excitatory synaptic transmission (Herring et al., 2013). However, genetic deletion of CNIH2 in hippocampal CA1 neurons strongly diminishes AMPAR-mediated synaptic transmission (Herring et al., 2013). In addition, a recent study has shown that in HEK cells, FRRS1L (C9orf4) can reduce non-saturating 100 µM glutamate-evoked, GluA1 homomermediated calcium influx, but increase GluA1-TARP γ8 complexmediated calcium influx (Erlenhardt et al., 2016). These data suggest that while FRRS1L does not change the number of receptors at the cell surface as shown in our study, it can modulate the steady-state currents in the absence of cyclothiazide and can functionally modulate AMPAR-TARP γ8 complexes in heterologous cells.
sgRNA-based single-cell KO of FRRS1L shows that FRRS1L plays an important role in the regulation of AMPAR-mediated synaptic transmission in hippocampal pyramidal neurons. Indeed, FRRS1L KO in hippocampal neurons both in vitro and in vivo reduces the AMPA EPSCs. Importantly, a sgRNA resistant FRRS1L fully rescues the AMPA EPSC deficit in the sgRNA-positive neurons. Similarly, FRRS1L KO in neuronal cultures significantly reduces GluA1 staining on the neuronal surface. These data are in agreement with a previous study in a neuronally differentiated cell line in which knockdown of FRRS1L diminished AMPA-induced whole-cell currents (Madeo et al., 2016). In addition, GluA1 total protein levels are reduced in FRRS1L KO neurons, suggesting that FRRS1L is important for GluA1 stability and/or biogenesis. Furthermore, the surface to total ratio of GluA1 expression is significantly decreased in neurons lacking FRRS1L. Thus, it is likely that decreased surface levels of GluA1 in FRRS1L KO neurons are caused by a combined effect of reduced total GluA1 expression and impaired delivery of GluA1 to the neuronal surface. It has been reported that expression levels of AMPAR receptor subunits are reduced in neurons lacking other AMPAR auxiliary subunits, including TARPs, CNIHs, PORCN or CKAMP44 (Rouach et al., 2005;Herring et al., 2013;Khodosevich et al., 2014;Erlenhardt et al., 2016), indicating that AMPAR auxiliary subunits commonly play important roles in the regulation of AMPAR stability in neurons.
While this manuscript was in preparation, an elegant study on FRRS1L that reveals a critical role of FRRS1L in early biogenesis of AMPARs and brain function was published (Brechet et al., 2017). This work shows that endogenous FRRS1L is largely confined in the ER, is important for AMPAR biogenesis and is required for AMPAR-mediated synaptic transmission (Brechet et al., 2017). Our data support these conclusions (Brechet et al., 2017). Indeed, similar to our single-cell KO data, it was reported that shRNA-mediated knockdown of FRRS1L diminished AMPAR-medicated synaptic transmission in hippocampal neurons (Brechet et al., 2017). In addition, our data show the reduced expression levels of GluA1 in FRRS1L KO neurons (Figure 5C), which is consistent with a role of FRRS1L in AMPAR biogenesis (Brechet et al., 2017). We also found that over-expression or single-cell KO of FRRS1L in hippocampal neurons did not change the AMPA EPSC decay kinetics, which is in agreement with the recent report (Brechet et al., 2017). Furthermore, our data show that FRRS1L, but not TARP γ8 or CNIH2, associates with dynein, indicating that FRRS1L-containing AMPARs represent a distinct receptor population, consistent with the reported proteomic analysis of the unique FRRS1L-AMPAR assemblies (Brechet et al., 2017). It has also been reported that over-expression of FRRS1L increases AMPAR-mediated synaptic transmission (Brechet et al., 2017), which is different from our findings that FRRS1L over-expression did not change AMPA EPSCs. This discrepancy can be explained by different over-expression techniques (viral transduction in vivo versus plasmid transfection in vitro in our study), different experimental preparations (acute rat hippocampal slices versus mouse hippocampal organotypic cultures in our study), different neuronal types for recording (hippocampal mossy cells, interneurons or CA3 neurons versus hippocampal CA1 neurons in our study) and different overexpression durations (∼ 2 weeks versus 2-5 days in our study). It is possible that over-expression of FRRS1L in neurons may not represent the appropriate approach to study the FRRS1L function in synaptic transmission, as over-expressed FRRS1L can reach the neuronal surface ( Figure 1F), but the vast majority of native FRRS1L do not (Brechet et al., 2017). Nevertheless, our findings of FRRS1L localization at dynein-positive vesicles together with the published report (Brechet et al., 2017) suggest that FRRS1L may primarily regulate the strength of AMPARmediated synaptic transmission through the modulation of AMPAR biogenesis and/or transport at the early secretory pathways.
In summary, we have provided some basic characterizations of FRRS1L in heterologous cells and hippocampal neurons. Our findings reveal an important role of FRRS1L in the regulation of excitatory synaptic transmission. Future work examining the molecular mechanisms underlying the regulation of AMPARs by FRRS1L and the potential role of dynein in the transport of FRRS1L-containing AMPARs will contribute to our understanding of AMPAR biogenesis and trafficking.

AUTHOR CONTRIBUTIONS
WH, HW, JL, SZ, and WL designed the research. HW performed immunocytochemical experiments in neurons, and electrophysiological recordings in HEK cells and hippocampal slice cultures. WH performed vesicle immunoisolation, immunocytochemical experiments in neuronal cultures and dual whole-cell current recording in acute hippocampal slices. WH and HW performed biochemical characterization. JL performed immunostaining in HEK cells. WL and WH wrote the paper. All authors read and commented on the manuscript, and approved the final version of the manuscript.