Local protein synthesis of neuronal MT1-MMP for agrin-induced presynaptic development

It has been widely acknowledged that protein synthesis occurs not only in the neuronal cell bodies, but also distally in the dendrites and axons (Biever et al., 2019; Holt et al., 2019; Spaulding and Burgess, 2017). In the last decades, local protein synthesis has been reported in different stages of neuronal development, ranging from growth cone guidance to synaptic plasticity (Dalla Costa et al., 2021; Kim and Jung, 2015). Recent comprehensive axonal translatome analyses have indicated that compartment-specific mRNA translation supports the formation and maintenance of neural circuits in vivo (Shigeoka et al., 2016). However, whether local protein synthesis is required for presynaptic differentiation at developing neuromuscular junctions (NMJs) remains largely unclear.

At vertebrate NMJs, presynaptic differentiation involves the clustering of mitochondria and synaptic vesicles (SVs), two of the most prominent presynaptic markers (Lee and Peng, 2006). Nerve-derived agrin not only regulates postsynaptic differentiation in skeletal muscles (McMahan et al., 1992), but serves as a stop signal to mediate presynaptic differentiation in ciliary ganglion neurons (Campagna et al., 1995; 1997). Our recent study reported that focal agrin stimulation directs vesicular trafficking and surface insertion of membrane-type 1 matrix metalloproteinase (MT1-MMP, also named MMP-14), which functions to proteolytically remodel the extracellular matrix environment and cleave full-length low-density lipoprotein receptor-related protein 4 (LRP4) for agrin deposition and signaling during presynaptic differentiation in motor neurons (Oentaryo et al., 2020). Therefore, the temporally delayed expression of surface MT1-MMP is believed to serve as a molecular switch from axonal outgrowth to synaptogenic phase in mature neurons. Considering the inefficiency of transporting MT1-MMP proteins synthesized from the soma of highly polarized mature neurons, it is of interest to determine whether MT1-MMP can be locally translated at the presynaptic terminals.

Using axon severing and microfluidic chamber assays, this study first shows that local synthesis of axonal proteins is required for agrin-induced assembly of presynaptic specializations in cultured Xenopus spinal neurons through mammalian target of rapamycin (mTOR)/eukaryotic initiation factor 4E-binding protein 1 (4E-BP1) signaling pathway. Live-cell time-lapse imaging coupled with photobleaching/photoconversion experiments demonstrate that ribonucleoprotein (RNP)-containing granules are locally captured by agrin stimulation to regulate MT1-MMP mRNA localization and translation. Collectively, this study identifies a novel requirement of MT1-MMP local translation in mediating agrin-induced presynaptic development.

To investigate the requirement of newly synthesized proteins in presynaptic differentiation, we first examined the effects of cycloheximide (CHX), a well-known protein synthesis inhibitor (Ennis and Lubin, 1964), on presynaptic differentiation in cultured Xenopus spinal neurons. Our recent study reported that agrin-coated beads induce structural and functional synaptic specializations along the neurites in a spatiotemporally controllable manner (Oentaryo et al., 2020). By adding 25 µM CHX to the cultured neurons before 4-h agrin bead stimulation, we found that agrin bead-induced mitochondrial and SV clustering were significantly inhibited (Fig. S1), indicating an essential requirement of newly synthesized proteins for the assembly of agrin-induced presynaptic specializations. Next, to determine whether the new proteins required for presynaptic differentiation are synthesized locally in axons, two parallel experimental approaches were adopted (Fig. 1A): (1) axons were severed and separated from the somas mechanically; and (2) axonal compartments were separated from the somas physically using microfluidic chambers. Consistent with the data above, CHX treatment in severed neurites or in the axonal compartment of microfluidic chambers also significantly inhibited agrin-induced clustering of mitochondria and SVs, as well as the localization of actin cytoskeletal scaffolds (F-actin), at the bead-neurite contacts (Fig. 1B-D). To visualize newly synthesized proteins, we used O-propargyl-puromycin (OPP), a click-reactive cell-permeable puromycin analog that can be incorporated into the translating peptides (Kim and Jung, 2015). We found that OPP signals, together with a presynaptic phosphoprotein synapsin I, were spatially localized at agrin bead-neurite contacts in control groups but were significantly inhibited by CHX treatment in either severed axons or the axonal compartment of microfluidic chambers (Fig. 1E-G). Together, these results indicate that local protein synthesis is required for the assembly of agrin-induced presynaptic specializations. It is important to note that the motor neuron is the major neuronal type in Xenopus cultures, as postsynaptic acetylcholine receptor clustering is detected at up to 90% of contacts between spinal neurons and muscle cells (Peng et al., 2003). Intriguingly, agrin is also required for presynaptic differentiation in hippocampal neurons (Bose et al., 2000), but not in pontine explant cultures (Scheiffele et al., 2000), which suggests the possible cell-type specificity of agrin in the induction of presynaptic differentiation.

We next investigated whether the well-studied mTOR signaling is involved in agrin-induced presynaptic development. Treatment with 0.2 µM rapamycin, an mTOR inhibitor (Tang et al., 2002), significantly inhibited agrin-induced formation of presynaptic specializations in either severed neurites or the axonal compartment of microfluidic chambers (Fig. 2A-C). Similar to CHX treatment, rapamycin treatment also effectively reduced the localized OPP and synapsin I signals at agrin bead-contacted sites (Fig. 2D-F).

Previous studies showed that mTOR promotes translation initiation through phosphorylating 4E-BP1 in cultured Xenopus neurons (Campbell and Holt, 2001; Piper et al., 2006). In our immunostaining experiments, p-mTOR and p-4E-BP1 signals were elevated at agrin bead-neurite contacts, whereas their total protein levels remained largely unchanged (Fig. 2G-H). However, rapamycin treatment in either severed neurites or the axonal compartment of microfluidic chambers greatly reduced p-mTOR and p-4E-BP1 signals at agrin bead-neurite contacts. As phosphorylated and total mTOR and 4E-BP1 proteins were probed by primary antibodies raised in the same species, we normalized each immunofluorescence marker intensity against a volume dye 5-(4,6-dichlorotriazinyl)-aminofluorescein (DTAF) (Schindelholz and Reber, 1999) to correct the differences in cell thickness. By performing such indirect ratiometric analyses, we demonstrated that rapamycin significantly reduced the phosphorylation of mTOR and 4E-BP1 at the agrin bead-neurite contacts (Fig. 2I). These findings collectively demonstrate that agrin promotes local protein synthesis required for presynaptic differentiation through mTOR/4E-BP1 signaling.

The axonal trafficking of mRNA is mediated by RNA-binding proteins and transported in RNP granules (Korsak et al., 2016; Sahoo et al., 2018). To visualize and track RNP granule movement in live neurons, we expressed fluorescent uridine triphosphate (Cy3-UTP) in cultured spinal neurons. Kymograph analyses showed agrin beads induce a stable localization of Cy3-UTP-labeled RNP granules along the neurite (Fig. 3A, arrow). By contrast, the percentage of bead-neurite contacts with localized Cy3-UTP signals and their intensities were significantly reduced by adding the functional blocking anti-agrin antibody or by using the negative control bovine serum albumin (BSA) beads (Fig. 3A,B).

To further demonstrate local capturing of RNP granules by agrin beads, we next performed fluorescence recovery after photobleaching (FRAP) experiments. Before photobleaching, stable Cy3-UTP-labeled RNP granules were detected at two of the agrin bead-neurite contacts (Fig. 3C). Cy3-UTP fluorescence at one of the contacts was photobleached (dotted region), followed by time-lapse imaging to track the dynamic movement of Cy3-UTP signals into the photobleached region. In this example, a moving Cy3-UTP granule was found to be transported to, and subsequently immobilized at, the agrin bead-contacted site as early as 2.6 s after photobleaching (arrowheads). By examining seven bead-neurite contacts from three independent experiments, we observed local capturing of the first moving Cy3-UTP granule by agrin beads 11.88±5.33 (mean±s.d.) seconds after photobleaching. Kymograph analyses further indicated the local capturing of multiple Cy3-UTP granules in the photobleached region over time. By contrast, the localized Cy3-UTP-labeled RNP granules at another agrin bead-neurite contact outside the photobleaching region appeared to be very stable throughout the entire imaging period (white arrows). Intriguingly, we found that the majority of bead-neurite contacts (58.29%) exhibited both spatially localized Cy3-UTP signals and mitochondrial clusters 30 min after agrin bead stimulation and was greatly abolished by CHX treatment (Fig. S2), which is consistent with other studies showing the inhibition of stress-induced RNP granule formation by CHX (Kedersha et al., 2000; Panas et al., 2016). Therefore, our results indicate that agrin-induced Cy3-UTP localization and mitochondrial clustering are temporally coupled events with very little time delay. Nevertheless, it is still plausible that RNP granule localization may be detected preceding the clustering of presynaptic markers shortly after agrin stimulation. Given the implications of mRNA localization and RNA-binding proteins in NMJ development (Bélanger et al., 2003; Chang et al., 2012), our study provides evidence showing the capturing and immobilization of RNP granules for mRNA localization and translation during agrin-induced presynaptic formation and maintenance.

Our recent study suggested that the delayed temporal expression of surface MT1-MMP serves as a molecular switch from axonal outgrowth to the synaptogenic phase in neuronal development (Oentaryo et al., 2020), indicating that MT1-MMP may be locally synthesized in axons of mature neurons. To test this, we performed single-molecule fluorescence in situ hybridization (smFISH) experiments that showed the spatial enrichment of MT1-MMP mRNA at agrin bead-neurite contacts (Fig. 4A,B). The specificity of MT1-MMP smFISH signals was validated by the treatment of RNase A and the addition of BSA beads, both of which largely abolished the localized signals at bead-neurite contact sites. Axonal RT-PCR analysis also revealed the presence of MT1-MMP mRNA in the pooled samples collected from both axonal and somal compartments (Fig. 4C), like the well-known axonal translation target β-actin mRNA (Zhang et al., 2001). Our results are consistent with a recent study demonstrating that MT1-MMP mRNAs are present in mouse forebrain synaptosomal transcriptome (Merkurjev et al., 2018), further supporting the possibility of MT1-MMP local protein synthesis at presynaptic regions.

To visualize the sites of MT1-MMP mRNA translation, we generated a plasmid of the photoconvertible Kaede cDNA (Leung and Holt, 2008) linked with the 3′ untranslated region (UTR) of MT1-MMP. Upon 10-s UV photoconversion, all existing green Kaede proteins were photoconverted into red proteins in severed neurites, which ruled out the somal contribution of newly synthesized Kaede proteins afterwards. After 30 min, we detected newly synthesized green Kaede proteins preferentially localized at the agrin bead-neurite contact in severed neurites (Fig. 4D,E), indicating that agrin induces local translation of the linked reporter protein via MT1-MMP 3′ UTR. It is known that the 3′ and/or 5′ UTR sequences are involved in synaptic targeting of mRNAs along the axons (Holt et al., 2019). However, the exact mechanisms underlying the axonal transport and synaptic targeting of MT1-MMP mRNA warrant further investigation.

To further demonstrate the local translation of MT1-MMP mRNA at sites of agrin stimulation, we next performed immunostaining experiments that showed the spatial enrichment of endogenous MT1-MMP proteins at the agrin bead-neurite contacts in either severed neurites or the axonal compartment of microfluidic chambers, but this spatial localization was largely reduced after CHX or rapamycin treatment (Fig. 4F-H). As expected, MT1-MMP antisense morpholino oligonucleotide, which effectively suppresses endogenous MT1-MMP expression by interfering with its translation initiation (Chan et al., 2020a; Oentaryo et al., 2020), also reduced MT1-MMP localization at the bead-neurite contacts (Fig. S3).

In summary, compared with the well-documented MT1-MMP intracellular trafficking mechanisms in neurons and other cell types (Chan et al., 2020b; Poincloux et al., 2009), this study reveals a novel regulatory mechanism involving local MT1-MMP translation via the mTOR/4E-BP1 signaling pathway in presynaptic development. Although recent studies have provided compelling evidence demonstrating local protein synthesis in axonal branching and maintenance of Xenopus retinal ganglion cells in vivo (Wong et al., 2017; Yoon et al., 2012), technological advances in enhancing mRNA probe sensitivity and its specific targeting to the presynaptic motor neurons will be required in the future to allow us to further understand the regulation of MT1-MMP mRNA localization and translation at developing NMJs in vivo.

To generate 3′ UTR (MT1-MMP)-Kaede reporter, Kaede (BamH1/EcoR1) was first amplified from pRSET-pr-Kaede (Addgene, 99228), and then subcloned into the empty pCS2+ plasmid using the following primers: forward, 5′-ACTGGTGGACAGCAAATGG-3′; and reverse, 5′-CTCGAATTCTTACTTGACGTTGTCC-3′. The 3′ UTR of Xenopus MT1-MMP was amplified with primers designed using the NCBI sequence (NM_001091009.1): forward, 5′-TAACCTCGAGATATGGAGCCTCTGAGAGCA-3′; and reverse, 5′-TACGCTCGAGAAAAAACTGAATTTGTTCATTTGAC-3′. The amplified fragment was then inserted into pCS2+ Kaede at the XhoI site.

The DNA construct encoding pCS2+3′ UTR (MT1-MMP)-Kaede was microinjected into the one-cell stage Xenopus embryos using an oocyte injector Nanoject (Drummond Scientific) as described previously (Lee et al., 2009, 2013). To visualize the localization of RNP granules, 13.8 nl Cy3-UTP (100 µM; ApexBio) was microinjected per embryo. To block the translation initiation of MT1-MMP mRNA, a custom-designed antisense morpholino oligonucleotide sequence against Xenopus MT1-MMP was used as reported previously (Oentaryo et al., 2020). Alexa Fluor 488-conjugated dextran (Thermo Fisher Scientific) was co-injected into the embryos as a fluorescent cell-lineage tracer.

Neural tubes were dissected from wild-type or microinjected embryos when they reached Nieuwkoop and Faber stage 19-22, and were then plated directly (for preparing neural explants) or treated with Ca²⁺/Mg²⁺-free solution before plating (for preparing dissociated neuron cultures) on acid-washed glass coverslips or microfluidic chambers coated with 71.4 µg/ml entactin-collagen IV-laminin (Merck Millipore). All experiments involving Xenopus frogs and embryos were carried out in accordance with the approved protocols of the Committee on the Use of Live Animals in Teaching and Research of The University of Hong Kong.

Cultured neurons were maintained in culture medium containing 10% Leibovitz's L-15 medium (v/v; Sigma-Aldrich), 87% Steinberg's solution [v/v; 60 mM NaCl, 0.67 mM KCl, 0.35 mM Ca(NO₃)₂, 0.83 mM MgSO₄ and 10 mM HEPES (pH 7.4)], 1% fetal bovine serum (v/v), 1% penicillin/streptomycin (v/v) and 1% gentamicin sulfate (v/v) at room temperature. To study local protein synthesis, the axons were severed by either removing the 1-day old explants or by separating the axonal from the somal compartments in microfluidic chambers. In experimental groups involving pharmacological treatment, CHX (ApexBio) or rapamycin (ApexBio) were added to the culture medium before neurite severing or 1 h before agrin bead stimulation. To locally induce presynaptic differentiation in cultured neurons or in severed neurites, 4.5 µm polystyrene beads (Polysciences) were coated with 100 µg/ml agrin (R&D Systems) according to the established protocol (Lee et al., 2009), or with 100 µg/ml BSA (Sigma-Aldrich) as the negative control. Notably, agrin or BSA beads were added into the neuronal cultures inside a laminar flow cabinet to avoid contamination. Before imaging experiments, the cultures were kept untouched for a minimum of 20 min to allow the beads to settle and make stable attachment with the neurites on the culture substrate. In most experiments, cells were fixed after 4-h bead stimulation, which has previously been demonstrated to induce the assembly of presynaptic specializations to the plateau level (Lee and Peng, 2006, 2008).

To label mitochondria, coverslips or microfluidic chambers containing the live cultured neurons or severed neurites were stained with MitoTracker Red CMXRos (50 nM; Thermo Fisher Scientific) for 5 min. To detect the newly synthesized proteins, a Click-iT Plus OPP Alexa Fluor 488 protein synthesis assay kit (Thermo Fisher Scientific) was used according to the manufacturer's protocol. Coverslips or microfluidic chambers were fixed with 4% paraformaldehyde for 15 min, followed by cell permeabilization with 0.1% Triton X-100 for 10 min. After blocking in 2% BSA for 2 h in room temperature, cells were stained with the primary antibody for 2 h at room temperature or overnight at 4°C. Primary antibodies used in this study included anti-synaptotagmin (1:200; mAb48; Developmental Studies Hybridoma Bank), anti-synapsin I (1:1000; A6442; Thermo Fisher Scientific), anti-mTOR (1:50; 7C-10; Cell Signaling Technology), anti-p-mTOR (1:100; ab109268; Abcam), anti-4E-BP1 (1:100; 9644S; Cell Signaling Technology), anti-p-4E-BP1 (1:100; 9451S; Cell Signaling Technology) and anti-MT1-MMP (1:200; MAB3328; Millipore). It was followed by staining with Alexa Fluor secondary antibodies (1:400; Thermo Fisher Scientific) for 45 min. To normalize the thickness of neurites, cultures were stained with 2.5 µM DTAF (Sigma-Aldrich) for 20 min and then washed three times with PBS before mounting. Coverslips were mounted on glass slides with fluoromount-G mounting medium (Thermo Fisher Scientific).

In microfluidic chambers, 0.1 mg/ml recombinant agrin protein (R&D Systems) was added to the axonal compartment on day 2 for 24 h. After adding Trizol reagent (Invitrogen) to the axonal compartment, axonal materials were then collected. Axonal and somal materials from 160 chambers (neural tube tissues from five embryos per chamber) were pooled from four independent experiments, followed by RNA extraction using RNAiso Plus (Takara) according to the manufacturer's protocol. Total RNAs (750 ng) of each sample were reverse transcribed by using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific). Somal (2 μl) and axonal (4 μl) cDNA samples were amplified by PCR of 40 cycles at 60°C annealing temperature using Phusion II High Fidelity DNA polymerase (Thermo Fisher Scientific) using a PCR thermal cycler (Takara). PCR primers for Xenopus MT1-MMP were designed based on the NCBI sequence (NM_001091009.1): forward, 5′-ATCTGCGGACACACACCTTA-3′; and reverse, 5′-GGCGTATCTCTTCCGTCTCA-3′. PCR primers for Xenopus β-actin and H4 were used as indicated in a previous study (Bellon et al., 2017): β-actin forward, 5′-CGTAAGGACCTCTATGCCAA-3′; β-actin reverse, 5′-TGCATTGATGACCATACAGTG-3′; H4 forward, 5′-GGCAAAGGAGGAAAAGGACT-3′, and H4 reverse, 5′-GAGAGCGTACACCACATCCA-3′.

Cultured Xenopus spinal neurons were fixed in 4% paraformaldehyde in PBS for 10 min, followed by permeabilization in 70% ethanol overnight at 4°C. Sequences of 48 unique custom-designed Stellaris RNA FISH probes (Biosearch Technologies) against Xenopus MT1-MMP mRNA are listed in Table S1. Hybridization of Quasar 670 dye-labeled probes was carried out overnight in the dark at 37°C in accordance with the manufacturer's instructions. As a negative control, the cultures were subjected to RNase A treatment (200 µg/ml) before hybridization.

In FRAP experiments, photobleaching was performed using an Axio TIRF unit fitted in a Nikon inverted microscope equipped with an oil immersion 100× NA 1.46 DIC objective lens. To obtain the baseline of fluorescence intensity before photobleaching, 11 images were taken at 0.2-s intervals on neurons expressing Cy3-UTP. Images were captured through Metamorph (Molecular Devices) using an Evolve 512 EMCCD camera (Photometrics). The regions of bead-neurite contact sites in Cy3-UTP-expressing neurons were chosen for photobleaching. Photobleaching was performed by using an OBIS laser line (561 nm) with 80% laser intensity, which was kept the same for all FRAP experiments. After photobleaching, 150 images were taken at 1-s intervals to visualize the axonal movement of Cy3-UTP-labeled RNP granules. To quantify the FRAP experiments, the relative fluorescence intensity in the photobleached area was first measured and then normalized by the mean fluorescence intensity of the same region in the images acquired before photobleaching. Kymographs of fluorescence signals along the neurites were constructed from time-lapse series using ImageJ (National Institutes of Health).

For Kaede photoconversion experiments, explants were plated next to the microgrooves of microfluidic chambers. After being cultured for 2 days, neurites were severed to isolate the axons. Agrin beads were added in the axonal compartment of microfluidic chambers for 2 h. Photoconversion was performed by UV exposure on agrin bead-contacted severed neurites expressing Kaede signals for 10 s. The neurites were imaged at both green and red channels in 30 min after UV photoconversion.

All other fluorescence imaging on live or fixed cells was acquired using an inverted wide-field fluorescence microscope (IX-83, Olympus), which was controlled via the imaging software Micro-Manager (Open Imaging). Images were taken using an ORCA-Flash 4.0 LT+ camera (Hamamatsu) with identical parameters between control and experimental groups. The percentage values were quantified by scoring 150 samples from three independent experiments, unless otherwise specified. For quantifying the fluorescence intensities, more than 50 samples from three independent experiments were collected and analyzed, unless otherwise specified. To quantify the normalized mean intensities of the presynaptic markers or OPP, the mean fluorescence signals at bead-neurite contacts were normalized with that at non-bead contact regions of the same neurites. In some experiments, immunostaining intensity of cytosolic protein markers was first normalized by DTAF signals to account for the difference in cell thickness. Data analyses were conducted using ImageJ, and the statistical analyses were performed using Prism 7 (GraphPad Software).

We thank Anna Tse for her technical support in some molecular biology experiments.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199000