A balance of protein synthesis and degradation is critical for the dynamic regulation and implementation of long-term memory storage. The role of the ubiquitin-proteasome system (UPS) in regulating the plasticity at potentiated synapses is well studied, but its roles in depressed synaptic populations remain elusive. In this study, we probed the possibility of regulating the UPS by inhibiting the proteasome function during the induction of protein synthesis-independent form of hippocampal long-term depression (early-LTD), an important component of synaptic plasticity. Here, we show that protein degradation is involved in early-LTD induction and interfering with this process facilitates early-LTD to late-LTD. We provide evidence here that under the circumstances of proteasome inhibition brain-derived neurotrophic factor is accumulated as plasticity-related protein and it drives the weakly depressed or potentiated synapses to associativity. Thus, UPS inhibition promotes LTD and establishes associativity between weakly depressed or potentiated synapses through the mechanisms of synaptic tagging/capture or cross-capture.
Experimental evidence about the role of protein synthesis during the processes of synaptic plasticity and long-term memory storage is overwhelming and well established (Davis and Squire 1984; Kandel 2001). However, evidence has recently begun to accumulate suggesting that protein degradation through the ubiquitin-proteasome system (UPS) is an equally important regulator of memory formation (Fonseca et al. 2006; Karpova et al. 2006; Hegde 2010; Jarome and Helmstetter 2013). In UPS, proteins to be degraded are marked by covalent linkage to a small protein called ubiquitin then degraded by a proteolytic complex, the proteasome (Hegde 2010). Proteolysis by UPS is a critical molecular mechanism that controls a plethora of functions in the nervous system, including protein kinases, synaptic proteins, transcription factors, and other molecules critical for synaptic plasticity and memory (Hegde 2010). In general, UPS operates by removing the inhibitory constraints involved in the establishment of synaptic plasticity, thus promoting long-term memory (Hegde 2010; Jarome and Helmstetter 2013).
Recent findings demonstrate that the induction of long-term potentiation (LTP), a cellular correlate of memory, leads to a rapid increase in the rate of protein synthesis and protein degradation via the proteasome system (Fonseca et al. 2006; Karpova et al. 2006). Proteasome inhibition enhances the induction of early-LTP by stabilizing the proteins locally translated from pre-existing mRNAs and impairs the maintenance of late-LTP by inhibiting transcription (Dong et al. 2008, 2014). Proteasome activity is differentially regulated in different neuronal compartments, for instance, restricted protein degradation can occur at certain synaptic compartments, while global degradation can affect all synaptic compartments within a neuron (Hegde 2004; Segref and Hoppe 2009). Protein degradation by the proteasome is also required for synaptic tagging/capture (STC) (Cai et al. 2010), a late-associative process of LTP and long-term depression (LTD) (Sajikumar and Frey 2004). During STC, a weak activation of synaptic population, either by a protein synthesis-independent early-LTP or early-LTD, sets a “synaptic tag.” This tag benefits by capturing the plasticity-related proteins (PRPs) synthesized from a nearby synaptic input which is activated by strong activity of late-LTP or late-LTD resulting in the consolidation of weak synapses (Redondo and Morris 2011). Different molecules were proposed as possible synaptic tags and PRPs (Redondo and Morris 2011). Specifically, the brain-derived neurotrophic factor (BDNF) and its receptor tropomyosin receptor kinase B (TrkB) are some of the prominent players which act as a PRP and tag, respectively (Lu et al. 2011; Sajikumar and Korte 2011).
The goal of this study was to investigate whether, and how the induction of LTD in the hippocampal CA1 region affects the balance between protein synthesis and degradation. We found that protein degradation occurs during the induction of early-LTD and its interference by means of proteasome inhibition results in the accumulation of PRPs which are sufficient for changing the threshold of the weakly activated synapses to a more stable state. We demonstrated for the first time that BDNF acts as a PRP for early-LTD during proteasomal inhibition and promotes the associative interactions such as STC and cross-capture (Sajikumar and Frey 2004).
Materials and Methods
A total of 184 acute hippocampal slices obtained from 92 Wistar rats (male, 6–7 weeks old) were used for electrophysiological recordings. All the procedures of hippocampal slice preparation were carried out in compliance with the guidelines from the Animal Committee on Ethics in the Care and Use of Laboratory Animals of TU-Braunschweig and National University of Singapore. Briefly, rats were anesthetized in a chamber filled with 100% CO2 and then decapitated immediately, after which the brains were quickly removed and cooled in 2–4°C artificial cerebrospinal fluid (aCSF). The right hippocampi were dissected and transverse hippocampal slices (400 µm) were prepared by using a manual tissue chopper (entire procedure took 2–3 min) and then incubated for 3 h in an interface chamber (Scientific System Design) which was continuously perfused with oxygenated aCSF at a flow rate of 0.8 mL/min. The aCSF contained the following (in mM): 124 NaCl, 4.9 KCl, 1.2 KH2PO4, 2.0 MgSO4, 2.0 CaCl2, 24.6 NaHCO3, 10 d-glucose, equilibrated with 95% O2–5% CO2 (32 L/h) (Sajikumar et al. 2005). To evoke field EPSP (fEPSP) from Schaffer collateral/commissural-CA1 synapses, 2 monopolar, lacquer-coated, stainless-steel, electrodes (5 MΩ; AM Systems) were positioned at an adequate distance within the stratum radiatum of the CA1 region for stimulating 2 independent synaptic inputs S1 and S2 that converge to one neuronal population (Fig. 1A). For recording fEPSP (measured as its initial slope function), one electrode (5 MΩ; AM Systems) was placed in the CA1 apical dendritic layer and signals were amplified by a differential amplifier (Model 1700, AM Systems). The signals were digitized using a CED 1401 analog-to-digital converter (Cambridge Electronic Design).
After the preincubation period, an input–output curve (afferent stimulation vs. fEPSP slope) was generated. Test stimulation intensity was adjusted to elicit fEPSP slope of 40% of the maximal fEPSP response for both synaptic inputs S1 and S2. Early-LTD was elicited by a “weak” low-frequency stimulation (WLFS) protocol consisting of 900 bursts delivered at 1 Hz (1 Hz, impulse duration 0.2 ms per half-wave, a total of 900 stimuli) (Sajikumar and Frey 2003). To induce early-LTP, a “weak” tetanization (WTET) protocol consisting of a single stimulus train of 21 pulses at 100 Hz (stimulus duration of 0.2 ms/polarity) was used. The slopes of the fEPSPs were monitored online. Four 0.2-Hz biphasic constant-current pulses (0.1 ms/polarity) were used for baseline recording and testing at each time point.
All the drugs were prepared as concentrated stock solutions either in dimethyl sulfoxide (DMSO) (MG132, lactacystin, anisomycin, emetine) or sterile PBS (TrkB/Fc) and were diluted in aCSF immediately before bath application. The final concentration of DMSO was always kept as 0.1%, a concentration that has no effect on basal synaptic transmission (Navakkode et al. 2005). Proteasome inhibitors MG132 (Tocris) and lactacystin (Tocris) were both used at a concentration of 5 µM. The final concentration of two protein synthesis inhibitors anisomycin (Tocris) and emetine (Tocris) were used at 25 and 20 µM, respectively. Recombinant human TrkB/Fc chimera (R&D Systems) was used at a concentration of 1 μg/mL. AIDA, (R,S)-1-aminoindan-1,5, dicarboxylic acid (Tocris), was used at a concentration of 500 µM dissolved in 0.1% DMSO (Li et al. 2014).
The average values of the slope function of the fEPSP (mV/ms) per time point were analyzed using the Wilcoxon signed-rank test (Wilcoxon test) when compared within the group, or the Mann–Whitney U-test (U-test) when data were compared between the groups; P < 0.05 was considered as statistically significant difference (Sajikumar et al. 2007).
Proteasome Inhibition Reinforces Early-LTD to Protein Synthesis-Dependent Late-LTD
It was reported earlier that proteasome inhibitors such as MG132 or lactacystin reduce the magnitude of LTP when applied during its induction (Fonseca et al. 2006; Karpova et al. 2006). The open question remains, how inhibition of the proteasome activity changes synaptic plasticity in general, e.g., in the processes of LTD and how fast the effects would affect its associative properties. Therefore, we investigated its role in early-LTD. In a control set of experiments, we induced early-LTD in the synaptic input S1 by the application of WLFS, which resulted in a transient form of LTD lasting at least 2 h (Fig. 1B, filled circles). A separate control pathway S2 (Fig. 1B, open circles) remained stable at baseline values. Statistically significant depression was observed in S1 for up to 140 min (Wilcoxon test, P = 0.007) and until 155 min (U-test, P = 0.04) after the induction of early-LTD. To investigate the role of proteasome activity during early-LTD, we bath-applied 5 μM MG132 (Fig. 1C) or lactacystin (Fig. 1D) during baseline fEPSP recordings 30 min before and after the application of WLFS. In the presence of the proteasome inhibitors, early-LTD in synaptic input S1 was transformed into late-LTD (Fig. 1C,D, filled circles), while the control input (Fig. 1C,D, open circles) remained stable at baseline levels for the entire experimental session. In both cases, statistically significant depression was maintained up to the end of the recording period after WLFS (Wilcoxon test, P = 0.01, U-test, P = 0.02). Separate control experiments were carried out to investigate whether a proteasome inhibitor has to be present during the application of WLFS to be effective in generating late-LTD, or it is sufficient to apply proteasome inhibitors at any time after or before the WLFS. Interestingly, MG132 or lactacystin application 30 min after WLFS for 1 h (Supplementary Fig. 1A,B) or MG132 for 1 h and a washout of 30 min before WLFS was unable to induce late-LTD (Supplementary Fig. 1C).
It was intriguing to us to investigate whether the reinforced late-LTD required protein synthesis. To elucidate this question, we co-applied 25 μM anisomycin (ANI, Fig. 1E,F) or 20 μM emetine (EME, Fig. 1G,H) together with MG132 or lactacystin. In both cases, protein synthesis inhibition prevented late-LTD in S1 (filled circles, Fig. 1E–H) without altering the control potentials in S2 (open circles, Fig. 1E–H). Statistically significant depression was observed up to 85 min (Wilcoxon test, P = 0.03) or up to 75 min (U-test, P = 0.03) in Figure 1E, up to 175 min (Wilcoxon test, P = 0.02) or up to 165 min (U-test, P = 0.04) in Figure 1F, up to 45 min (Wilcoxon test, P = 0.02), or up to 60 min (U-test, P = 0.04) in Figure 1G, up to 80 min (Wilcoxon test, P = 0.04) or up to 65 min (U-test, P = 0.003) in Figure 1H.
Taken together, proteasome inhibition during early-LTD induction leads to a protein synthesis-dependent late-LTD.
Proteasome Inhibition Reinforces Early-LTD to Late-LTD and It Expresses Synaptic Tagging/Capture
Since proteasome inhibition fosters the generation of protein synthesis-dependent late-LTD, we investigated the possible candidate PRPs which were rescued from degradation during proteasome inhibition. In this context, we first tested if BDNF could be a PRP affected by proteasome inhibition, because we and others have reported earlier that BDNF acts as a critical PRP during the persistence of LTP and LTD (Lu et al. 2011; Sajikumar and Korte 2011). To test the role of BDNF, we co-applied recombinant human TrkB/Fc chimera (TrkB/Fc; 1 μg/mL) together with MG132 or with lactacystin 30 min before and after the induction of early-LTD (Fig. 2A,B). TrkB/Fc scavenges BDNF away and therefore blocks TrkB activation (Shelton et al. 1995). The reinforcement effect by proteasome inhibition was completely abolished in the presence of TrkB/Fc, indicating the role of BDNF as a PRP that was rescued from degradation during proteasome inhibition (Fig. 2A,B, filled circles). In Figure 2A, a statistically significant depression was evident in S1 up to 180 min (Wilcoxon test, P = 0.03) and only up to 120 min (U-test, P = 0.03). In Figure 2B, a statistically significant depression was maintained only up to 60 min (Wilcoxon test, P = 0.03, U-test, P = 0.02). A slight depression was obtained after the application of WLFS in S2 in Figure 2A that lasted up to 50 min but was not statistically significant at any time points.
Since proteasome inhibited-LTD (Pi-LTD) was maintained by BDNF, we explored the possible associative interactions such as STC. For investigating the processes of STC during Pi-LTD, early-LTD was induced in the synaptic input S1 in the presence of MG132 or lactacystin (Fig. 2C,D, drug application time of 30 min before and after WLFS) and after 30 min washout of the drug, WLFS was applied to S2, now without the presence of a proteasome inhibitor. Intriguingly, in both inputs, a late form of LTD was observed (Fig. 2C,D, filled and open circles), which supports our hypothesis that STC takes place during Pi-LTD. S1 and S2 of Figure 2C,D expressed statistically significant depression after the application of WLFS and maintained throughout the recoding period (Wilcoxon test, P = 0.01, U-test, P = 0.01). In a similar manner, continuous inhibition of BDNF by TrKB/Fc 1 h after the establishment of STC prevented not only the persistence of LTD in S1, but also the capture process in S2 providing further evidence for the role of BDNF as a critical PRP in the processes of maintaining STC (Sajikumar and Korte 2011) induced by proteasome inhibition (Fig. 2E,F, filled and open circles). Statistically significant depression was observed in S1 and S2 up to 220 min (Wilcoxon test, P = 0.02, U-test, P = 0.03) in Figure 2E and up to 225 min in Figure 2F (Wilcoxon test, P = 0.01, U-test, P = 0.02).
Cross-Capture by Proteasome-Inhibited LTD
Next, we asked whether a tag set due to the induction of a transient early-LTP in one synaptic input could benefit from the BDNF generated due to the Pi-LTD in another synaptic input. This positive associative interaction is defined as cross-capture (Sajikumar and Frey 2004). If cross-capture exists, then early-LTP could be transformed into late-LTP. To test this possibility, we first induced early-LTD in the presence of MG132 or lactacystin (30 min before and after the application of WLFS) in S1 (Fig. 3A,B, filled circles); 60 min later, i.e., 30 min after the washout of the drug, WTET was applied to S2 for inducing early-LTP (Fig. 3A,B, open circles). Surprisingly, the early-LTP in synaptic input S2 was transformed into late-LTP. Statistically significant depression or potentiation was observed in S1 and S2, respectively, in Figure 3A (Wilcoxon test, P = 0.007) and B (Wilcoxon test, P = 0.017) up to 240 min. This indicates that Pi-LTD was capable of providing PRPs for the consolidation of early-LTP in S2 and by this means promoting associativity. We also tested whether the reinforcement of early-LTP in S2 was due to the possible synthesis of BDNF during WLFS in S1. To rule out this possibility, we repeated the experiments depicted in Figure 3A,B but without proteasome inhibition during early-LTD induction and observed no cross-capture (Supplementary Fig. 2A). In a separate control experiment, we confirmed that prior application of MG132 for 1 h during baseline recording, similar to that of Figure 3A,B (open circles), and induction of early-LTP 30 min after the washout of the drug did not result in the reinforcement of early-LTP to late-LTP, confirming that the PRPs for early-LTP consolidation were captured from the PRPs produced by Pi-LTD (Supplementary Fig. 2B).
Given that BDNF plays a critical role in maintaining Pi-LTD, we asked whether BDNF can also function as a PRP for cross-capture in this context. To address this, we used the same experimental design as in Figure 3A,B but in addition TrkB/Fc (1 μg/mL) was bath-applied 120 min after the establishment of cross-capture (Fig. 3C,D). Interestingly, the maintenance of Pi-LTD in S1 (Fig. 3C,D, filled circles) and the transformation of early-LTP to late-LTP in S2 (Fig. 3C,D, open circles) were prevented. Significant depression was observed up to 180 min in S1 in Figure 3C (Wilcoxon test, P = 0.03), whereas a statistically significant potentiation was maintained up to 200 min in S2 (Wilcoxon test, P = 0.04). In Figure 3D, statistically significant depression and potentiation was maintained in S1 and S2 up to 205 and 200 min, respectively (Wilcoxon test, P = 0.04). Thus, cross-capture initiated by Pi-LTD was a BDNF-dependent process.
The ubiquitin-proteasome pathway regulates a plethora of synaptic functions in vertebrates and nonvertebrates (Hegde and DiAntonio 2002; Hegde 2010). More specifically, at mammalian synapses, long-term alterations in synaptic activity results in global changes in the composition of postsynaptic proteins, mainly through the UPS (Ehlers 2003). It has been reported earlier that two structurally distinct proteasome inhibitors such as MG132 and lactacystin block the degradation of the postsynaptic density protein, PSD-95, indicating that proteasome-dependent regulation is critical for LTD (Colledge et al. 2003). In contrast to this, our results suggest that inhibition of proteasome function during the induction of a protein synthesis-independent form of early-LTD has a positive effect, resulting in the generation of protein synthesis-dependent late-LTD (Supplementary Fig. 3). These results further highlight the importance of protein degradation during the early-phase of LTD which is believed to be independent of protein synthesis. Overall, the balance between protein synthesis and degradation seems to be critical for early-LTD similar to LTP (Fonseca et al. 2006). Our results are in line with recent findings that proteasome inhibition by MG132 facilitates LTD through STriatal-Enriched protein tyrosine Phosphatase (STEP) (Chen et al. 2013). STEP is a brain-specific tyrosine phosphatase that modulates N-methyl-d-aspartate receptor (NMDAR) trafficking which then regulates LTP (Snyder et al. 2005; Braithwaite et al. 2006). Earlier studies convincingly showed that administration of STEP into hippocampal slices prevented the induction of LTP, and STEP knockout mice show enhanced LTP (Pelkey et al. 2002; Zhang et al. 2010). STEP dephosphorylates a regulatory Tyr within the activation loop of several kinases resulting in the inactivation of the respective protein. One of the important kinase to mention here is extracellular signal-regulated kinase 1/2 (ERK1/2), since we have reported earlier that ERK1/2 plays an important role in the conversion of early to late-LTD in a phosphodiesterase inhibited situation in hippocampal CA1 pyramidal neurons (Navakkode et al. 2005). Given the importance of STEP in establishing LTD, and the role of proteasome inhibitors such as MG132 in activating STEP, we hypothesized that proteasome inhibition will lead to mGluR-mediated LTD. This is supported by our finding that mGluR antagonist completely abolishes the facilitating effects of proteasome-reinforced LTD (Supplementary Fig. 4). Therefore, it is interesting to speculate, that we might have directly activated STEP using proteasome inhibitors which then resulted in the protein synthesis-dependent facilitation of early-LTD to late-LTD.
The two forms of LTD in CA1 pyramidal neurons, NMDAR-LTD, and mGluR-LTD, are differentially regulated by the UPS (Citri et al. 2009). It has been reported that NMDAR-induced AMPAR endocytosis and LTD occur independently of proteasome function, but mGluR-induced AMPAR endocytosis and LTD are enhanced by inhibition of proteasomal degradation by a mechanism that involves the UPS (Citri et al. 2009). Mao et al. (2009) reported that interfering with proteasome activity during NMDAR-LTD in the nucleus accumbens also yields a similar effect as it was observed during proteasome inhibition in hippocampal mGluR-LTD. Our findings are in line with these two observations and provide additional mechanistic insights.
Fonseca et al. reported earlier that, during LTP, inhibition of the proteasome activity results in the accumulation of “negative” effector proteins, which would normally be targeted for degradation after LTP induction, thereby destabilizing the balance of protein synthesis and degradation resulting in the impairments of LTP (Fonseca et al. 2006). Our results now significantly advance this important observation by showing that proteasome inhibition during early-LTD induction leads to the generation of BDNF as a PRP. Interestingly, Dong et al reported an enhancement of early-LTP during proteasome inhibition and noted that the observed effect was due to the translation of pre-existing mRNAs (Dong et al. 2008) (here BDNF mRNA) in dendrites. Further evidence also suggest that BDNF can act as a PRP in late-LTP in a transcription-independent manner strengthening our hypothesis that locally synthesized PRPs like BDNF can establish long-term plasticity and associativity (Barco et al. 2005; Sajikumar and Korte 2011).
It has been reported earlier that BDNF and its receptor TrkB can act as a possible PRP and synaptic tag, respectively (Lu et al. 2011; Sajikumar and Korte 2011). TrkB is a membrane protein, which can be activated in a local and synapse-specific manner. TrkB activation is also confined to dendritic spots near the site of local BDNF secretion and it tends to be associated with the cell membrane after its secretion, with very limited diffusion (Nagappan et al. 2009; Lu et al. 2011). We favor this model in the light of our previous and present findings that BDNF can be tagged specifically to depressed synapses, which enables long-term plasticity and associativity (Sajikumar and Korte 2011). In the present study, we observed that early-LTP in the presence of proteasome inhibition is mGluR-dependent and capture of BDNF is critical not only for LTD tagging but also for the conversion of early-LTP to late-LTP in cross-capture. We do not exclude the possibility of the activation of multiple mechanisms during the processing of information within the same neuronal compartments having different plasticity thresholds (Sajikumar and Korte 2011). The reversal of LTD by BDNF scavenger is analogous to the reversal of the maintenance of LTP by protein kinase Mzeta (PKMζ) inhibitors reported in the earlier studies (Serrano et al. 2005). Interestingly, the only other agent known to reverse LTP other than PKMζ inhibitors is the same TrkB/Fc, which reverses theta-burst stimulation-induced potentiation in a time window within 2-h poststimulation (Huang and Kandel 1995; Sajikumar and Korte 2011). However, the possibility of the role of PKMζ as a critical plasticity protein that establishes Pi-LTD in our present findings is less likely because of two reasons. 1) PKMζ mainly acts as a PRP for potentiated synapses and not for the depressed synapses (Sajikumar et al. 2005). 2) If a second PRP-like PKMζ is present, one would still expect cross-capture in the presence of TrkB/Fc maintained by PKMζ similar to that of the findings described in Sajikumar and Korte (2011). The WLFS used in our study might have resulted in the synthesis of BDNF locally or used translocated BDNF from other synaptic sides during proteasome inhibition which acted as a common PRP for both LTD and LTP during the conversion of weakly potentiated synapses during cross-capture. This observation highlights an important aspect of the tag-PRP interaction in tagging and cross-capture: BDNF-dependent tagging is a local dynamic process and in a global situation, for instance, in case of metaplastic regulation of tagging and capture, tag-PRP interaction must be regulated by multiple PRPs (Sajikumar and Korte 2011). In short, the present findings show a specific role of BDNF as a PRP during Pi-LTD and a PRP for establishing associativity in depressed and potentiated synapses, respectively.
Since proteasome function is critical for the processes of synaptic plasticity, an important cellular correlate of memory formation, more studies are required to probe whether the balance between protein synthesis and proteasome function can be explored in order to restore plasticity in neurodegenerative diseases. This is of critical importance for future clinical studies since the accumulation of protease-resistant misfolded and aggregated proteins is a common mechanism underlying protein misfolding disorders like Huntington's disease, Alzheimer's disease, Parkinson's disease, prion diseases, and amyotrophic lateral sclerosis (Ciechanover and Kwon 2015).
This work was supported by the Deutsche Forschungsgemeinschaft (SA 1853/1-1; to S.S. and M.K.). S.S. was additionally supported by an Alexander von Humboldt Fellowship and National Medical Research Council Collaborative Research Grant (NMRC-CBRG-0041/2013). Q.L. was supported by DAAD fellowship (A/09/98265).
We are grateful to Dr Sheeja Navakkode, Mahima Sharma, and Mahesh Shivarama Shetty for their helpful discussions and Reinhard Huwe for his outstanding technical assistance. Conflict of Interest: None declared.