|Home | About | Journals | Submit | Contact Us | Français|
Beta-amyloid precursor protein cleavage enzyme 1 (BACE1) has been identified as a major neuronal β-secretase critical for the formation of β-amyloid (Aβ) peptide, which is thought responsible for the pathology of Alzheimer’s disease (AD). Therefore, BACE1 is one of the key therapeutic targets that can prevent the progression of AD. Previous studies showed that knocking out the BACE1 gene prevents Aβ formation, but results in behavioral deficits and specific synaptic dysfunctions at Schaffer collateral to CA1 synapses. However, BACE1 protein is most highly expressed at the mossy fiber projections in CA3. Here we report that BACE1 knockout mice display reduced presynaptic function, as measured by an increase in paired-pulse facilitation ratio. More dramatically, mossy fiber LTP, which is normally expressed via an increase in presynaptic release, was eliminated in the knockouts. While LTD was slightly larger in the BACE1 knockouts, it could not be reversed. The specific deficit in mossy fiber LTP was upstream of cAMP signaling, and could be “rescued” by transiently elevating extracellular Ca2+ concentration. These results suggest that BACE1 may play a critical role in regulating presynaptic function, especially activity-dependent strengthening of presynaptic release, at mossy fiber synapses.
Alzheimer’s disease (AD) is the most prevalent form of senile dementia. Current treatment of AD remains limited, and there is no effective disease-modifying treatment as of yet (Citron, 2004b). It is widely believed that AD is initiated as a synaptic dysfunction, which correlates with the loss of memory function in the early stages of the disease (Selkoe, 2002). A current hypothesis states that over-production of amyloid-beta (Aβ) peptide initiates the pathogenesis of AD (Hardy and Selkoe, 2002; Citron, 2004b; Walsh and Selkoe, 2007). Aβ is produced by the sequential cleavage of amyloid precursor proteins (APPs) by β- and γ-secretases, which are one of the major disease-modifying targets to treat AD (Citron, 2004b). However, it became apparent that γ-secretase processes other critical substrates essential for normal cell development and function, such as Notch (Sisodia and St George-Hyslop, 2002; Selkoe and Kopan, 2003). Therefore, inhibiting β-secretase is now receiving renewed attention (Vassar, 2002; Citron, 2004b, 2004a). The amount and activity of β-secretase is elevated in sporadic AD brains (Yang et al., 2003; Li et al., 2004; Zhao et al., 2007), further suggesting that effective methods to reduce its activity may be beneficial to a large population of AD patients.
A transmembrane aspartic protease, called BACE1 (beta-site APP cleavage enzyme 1), was identified as the major neuronal β-secretase (Hussain et al., 1999; Sinha et al., 1999; Vassar et al., 1999; Yan et al., 1999). BACE1 knockout mice were generated to determine the functional consequences of chronically inhibiting the activity of β-secretase. Initial characterization of the BACE1 knockouts suggested that there are no gross anatomical or functional abnormalities (Luo et al., 2001; Luo et al., 2003). Moreover, knocking out BACE1 in APP transgenic lines, which normally develop Aβ plaques and behavioral deficits, essentially alleviated the AD symptoms (Luo et al., 2003; Ohno et al., 2004; Laird et al., 2005). However, recent studies, including our own, showed that BACE1 knockouts display specific dysfunctions in synaptic transmission and plasticity (Ohno et al., 2004; Laird et al., 2005), as well as behavioral deficits (Harrison et al., 2003; Laird et al., 2005; Savonenko et al., 2008). While all of the studies characterizing synaptic function of BACE1 knockouts so far have been carried out in the CA1 region of the hippocampus (Ohno et al., 2004; Laird et al., 2005; Ma et al., 2007), the expression of BACE1 is most prominent in the mossy fiber terminals that synapse onto CA3 pyramidal neurons (Laird et al., 2005; Zhao et al., 2007). Therefore, we examined synaptic function and plasticity of the BACE1 knockouts at the mossy fiber synapses.
All mice used (BACE1 +/+ and −/−) were derived from heterozygous breeders (+/−) as described previously (Laird et al., 2005). The Institutional Animal Care and Use Committees of both University of Maryland at College Park and Johns Hopkins University approved all procedures involving animals.
Hippocampal slices (400-μm-thick) were prepared from adult (3–6 months old) male BACE1 knockout and wildtype mice as previously described (Laird et al., 2005). Briefly, hippocampi were sliced in ice-cold dissection buffer (in mM: 212.7 sucrose, 2.6 KCl, 1.23 NaH2PO4, 26 NaHCO3, 10 dextrose, MgCl2, and 1 CaCl2; saturated with 5% CO2 and 95% O2). Recordings were done in a submersion-type recording chamber perfused with artificial cerebrospinal fluid (ACSF, in mM: 124 NaCl, 5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 10 dextrose, 1.5 MgCl2, and 2.5 CaCl2; saturated with 5% CO2/95% O2; 29.5°C–30.5°C, 2 ml/min). Synaptic responses were evoked through bipolar stimulating electrodes (double-barreled borosilicate glass capillaries [Sutter Instruments, Novato, CA]) placed in the dentate granule cell layer to activate the mossy fibers with pulse durations of 0.2 ms (baseline stimulation at 0.067 Hz), and recorded extracellularly in the stratum lucidum of CA3. Both the stimulating and recording electrodes were filled with ACSF. To induce long-term potentiation (LTP), three trains of 100 Hz (1 sec) stimuli were given at 20 sec intervals. Long-term depression (LTD) was induced by a paired-pulse 1Hz protocol (interstimulus interval (ISI) = 50 ms, 15min). For measurement of paired-pulse facilitation (PPF), ISIs of 25, 50, 100, 200, 400, 1000, and 2000 ms were used. In some experiments, extracellular Ca2+ concentration was increased to 5.0 mM for 10 minutes before delivering HFS (Castillo et al., 2002). To activate cAMP production, 50 μM forskolin (Sigma-Aldrich, St. Louis, MO) was applied for 5 minutes. All experiments were done in the presence of 100 μM D,L-2-amino-5-phosphonovaleric acid (D,L-APV) (Sigma-Aldrich, St. Louis, MO) to block N-methyl-D-aspartate (NMDA) receptors. At the end of each experiment, 1 μM (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine (DCG-IV) (Tocris Bioscience, Ellisville, MI) was added, and blockade ≥ 80% were taken to be mossy fiber inputs. Field potential slopes were measured, and data are expressed as mean ± standard error of mean.
We previously observed that mossy fiber terminals are enriched in BACE1 protein compared to other hippocampal subfields (Laird et al., 2005). Therefore, we hypothesized that BACE1 knockouts may exhibit alterations in synaptic transmission and plasticity at this particular set of synapses. We first measured presynaptic function by comparing paired-pulse facilitation (PPF) ratio at various interstimulus intervals (ISIs). We found a significant interaction between the genotype and ISIs (Two-factor ANOVA, genotype * ISI: F(6, 203) = 2.586, P < 0.02), particularly BACE1 KOs displayed larger PPF ratios at shorter ISIs (25 ms ISI: WT = 3.4 ± 0.57; KO = 6.1 ± 0.79; 50 ms ISI: WT = 3.6 ± 0.65, n = 14; KO = 5.7 ± 0.77, n = 17; Fisher’s PLSD posthoc test: P < 0.002 for 25 and 50 ms ISI between WT and KO; Fig. 1A). The increase in PPF ratio suggests a reduction in presynaptic release. Synaptic transmission at mossy fiber to CA3 synapses display sensitivity to group II mGluR agonists (Nicoll and Schmitz, 2005). Bath application of 1 μM DCG-IV at the end of the recording produced a comparable reduction in basal synaptic transmission in both knockouts and wildtypes (WT: 12 ± 4% of baseline at 20 min DCG-IV, n = 14; KO: 11 ± 2%, n = 17; Fig. 1B).
Next we examined whether knocking out BACE1 affects synaptic plasticity at the mossy fiber synapses. We first compared LTP induced by high frequency stimulation (HFS: 3 × 100 Hz, 1 sec). BACE1 knockouts showed a larger initial potentiation, suggesting an enhanced facilitation during HFS, however, the responses relaxed back to baseline by 1 hour (WT: 149 ± 10% of baseline at 1 hour post-HFS, n = 13 slices/6 mice; KO: 96 ± 7%, n = 16 slices/7 mice; t-test: P < 0.01; Fig. 2A). Consistent with a presynaptic locus of expression, LTP in wildtypes was accompanied by a decrease in PPF ratio measured at 50 ms ISI (baseline = 3.1 ± 0.5, 1 hour post-HFS = 2.6 ± 0.4, n = 13 slices/6 mice, paired t-test: P < 0.03), but knockouts displayed a trend of a decrease in PPF ratio which returned to basal levels at 1 hour (baseline = 5.9 ± 1.0, 20 min post-HFS = 3.6 ± 0.5, 1 hour post-HFS = 6.1 ± 1.2, n = 16 slices/7 mice, ANOVA: F(2,45) = 2.018, P = 0.1).
To test LTD, we used a paired-pulse 1 Hz protocol (PP-1 Hz, 15 min), because a standard 1 Hz (15 min) protocol (Kobayashi et al., 1996) failed to produce LTD in the wildtypes at the ages used for our study (data not shown). LTD induced by the PP-1 Hz was slightly, but significantly, larger in BACE1 knockouts (WT: 75 ± 4.3% of baseline at 1 hour post-onset of PP-1 Hz, n = 7 slices/5 mice; KO: 62 ± 3.8% of baseline, n = 6 slices/4 mice; t-test: P < 0.04; Fig. 2B). This form of LTD did not significantly change PPF ratio either in wildtypes or knockouts (WT: baseline = 3.8 ± 1.0, 1 hour post-PP 1 Hz = 3.2 ± 0.9, n = 6 slices/4 mice, paired t-test: P = 0.16; KO: baseline = 7.1 ± 1.5, 1 hour post-PP 1Hz = 5.5 ± 1.1, n = 6 slices/4 mice, paired t-test: P = 0.10). Unlike in wildtypes, HFS failed to reverse LTD in the knockouts (WT: 195 ± 28.0% of renormalized baseline at 1 hour post-HFS, n = 6 slices/4 mice; KO: 100 ± 5.4%, n = 6 slices/4 mice; t-test: P < 0.02; Fig. 2B).
Mossy fiber LTP is triggered by a rise in presynaptic Ca2+(Castillo et al., 1994), and a further recruitment of cAMP-dependent signaling mechanisms (Nicoll and Schmitz, 2005). Therefore, we investigated whether the lack of mossy fiber LTP in BACE1 knockouts is due to abnormal regulation of presynaptic Ca2+ or signaling downstream. We found that transiently increasing the concentration of extracellular Ca2+ (from 2.5 mM to 5 mM) during HFS recovered mossy fiber LTP in BACE1 knockouts (137 ± 7.9% of baseline at 1 hour post-HFS, n = 9 slices/6 mice; paired t-test: P < 0.01; Fig. 3A). Furthermore, LTP was accompanied by a decrease in PPF ratio measured at 50 ms ISI (baseline = 4.2 ± 0.7, 1 hour post-HFS = 2.4 ± 0.4, n = 9 slices/6 mice, paired t-test: P < 0.01) consistent with a presynaptic expression. Increasing external Ca2+ concentration alone produced only a transient potentiation (110 ± 4.8% of baseline at 1 hour post-Ca2+, n = 4 slices/2 mice; paired t-test: P = 0.13; Fig. 3A).
To further confirm whether signaling downstream of the Ca2+ signal is intact in BACE1 knockouts, we directly activated cAMP signaling by a brief application of an adenylyl cyclase activator forskolin. This caused a dramatic enhancement of synaptic transmission in both wildtypes and knockouts to similar magnitudes (WT: 622.5 ± 57.8% of baseline at 1 hour post-forskolin, n = 7 slices/5 mice; KO: 741.8 ± 110.1%, n = 7 slices/4 mice; t-test: P = 0.36; Fig. 3B). This was accompanied by a significant decrease in PPF ratio in both genotypes (WT: baseline = 3.2 ± 0.34, 1 hour post-forskolin = 1.5 ± 0.15, n = 7 slices/5 mice, paired t-test: P < 0.01; KO: baseline = 4.9 ± 0.63, 1 hour post-forskolin = 1.7 ± 0.16, n = 7 slices/4 mice, paired t-test: P < 0.01), consistent with a presynaptic mechanism of potentiation. This demonstrates that the presynaptic deficits seen in BACE1 knockouts are upstream of cAMP signaling.
We found that BACE1 knockouts display severe deficits in presynaptic function at mossy fiber synapses in CA3: a reduction in presynaptic release and an absence of mossy fiber LTP. In addition, BACE1 knockouts exhibited a slightly larger LTD, which could not be reversed. These results suggest that BACE1 function is critical for normal synaptic transmission and plasticity, especially activity-dependent potentiation, at these synapses. We further found that the specific deficit in mossy fiber LTP in BACE1 knockouts can be rescued by increasing extracellular Ca2+ concentration. Since a direct activation of cAMP production was not impaired in the BACE1 knockouts, our data suggests that the presynaptic dysfunction is likely at the level of presynaptic Ca2+ regulation.
Previous studies suggest that BACE1 is highly localized to presynaptic terminals, especially at the mossy fiber boutons in the CA3 (Laird et al., 2005; Zhao et al., 2007). This localization is consistent with our observation of a deficit in presynaptic function and plasticity at this synapse. Taken together with our previous results from the CA1 also showing an increase in PPF ratio (Laird et al., 2005), these results indicate that BACE1 may play a general role in regulating presynaptic function under physiological conditions. However, whether presynaptic deficits in BACE1 knockouts are directly due to lacking APP processing is unclear. Previous studies suggest that generation of excess Aβ depresses excitatory synaptic transmission mainly by postsynaptic removal of AMPA receptors and loss of synapses (Hsieh et al., 2006; Priller et al., 2006; Ting et al., 2007). These results would predict that lacking Aβ production, as in BACE1 knockouts, would cause a postsynaptic increases in AMPA receptor function, not a decrease in presynaptic function as observed in our studies. However, we cannot rule out the possibility of gain-of-function in the Aβ over-expression studies.
Another possibility is that the presynaptic effects of BACE1 knockout may be from abnormal processing of substrates other than APP. It is now known that BACE1 can also cleave APP-like proteins (APLPs) (Li and Sudhof, 2004), β subunits of voltage-gated Na+ channel (Wong et al., 2005; Kim et al., 2007), and neuregulin-1 (NRG1) (Hu et al., 2006; Willem et al., 2006). Regulation of the latter two substrates is particularly interesting. The β2 subunit of Na+ channel is critical for plasma membrane expression of functional Na+ channels (Schmidt and Catterall, 1986), which are essential for action potential generation. However, over-expressing BACE1 actually decreases the density of functional Na+ channels (Kim et al., 2007), hence it cannot directly account for the observed reduction in presynaptic release in the BACE1 knockouts. Potential regulation of NRG1 by BACE1 was discovered from observations that BACE1 knockouts display a hypomyelination phenotype with a correlated accumulation of full-length NRG1 and a significant loss of NRG1 cleavage products (Hu et al., 2006; Willem et al., 2006). Recently, we demonstrated that the lack of NRG1 processing in BACE1 knockouts reduces postsynaptic function of ErbB4, a receptor for NRG1 (Savonenko et al., 2008). NRG1/ErbB4 signaling has been suggested to regulate synaptic function and plasticity, mainly via regulation of postsynaptic glutamate receptors (Huang et al., 2000; Gu et al., 2005; Li et al., 2007). Nevertheless, abnormal processing of NRG1 may also affect presynaptic release by regulating the expression of nicotinic acetylcholine receptor (nAchR) subunit a7 (Liu et al., 2001), which allows Ca2+ influx (Seguela et al., 1993). Indeed, presynaptic nAchRs can increase glutamate release (McGehee et al., 1995; Gray et al., 1996; Maggi et al., 2003), likely via the α7 containing nAchRs (Le Magueresse et al., 2006). These results suggest that lacking NRG1 cleavage, as in BACE1 knockouts, would reduce presynaptic release. Whether this is the case for mossy fiber synapses is unclear (Vogt and Regehr, 2001).
Our results indicate that a complete inhibition of BACE1 activity is deleterious for neuronal function, especially at the mossy fiber synapses in CA3 compared to Schaffer collateral inputs in CA1. This suggests that mossy fiber dysfunction may have had a larger impact on the behavioral phenotypes seen in the BACE1 knockouts (Harrison et al., 2003; Laird et al., 2005; Savonenko et al., 2008). We demonstrate that signaling downstream of presynaptic Ca2+ is intact in BACE1 knockouts. Therefore, we were able to restore mossy fiber LTP in the BACE1 knockouts by simply increasing extracellular Ca2+ concentration during LTP induction. This has significant clinical implications, because it suggests that means to enhance presynaptic Ca2+ will circumvent synaptic deficits, and perhaps alleviate the behavioral phenotypes, associated with inhibiting BACE1 activity.
This work was supported by a grant from the National Institute of Health (P01-NS047308).