MMP13 inhibition rescues cognitive decline in Alzheimer transgenic mice via BACE1 regulation

Abstract

MMP13 (matrix metallopeptidase 13) plays a key role in bone metabolism and cancer development, but has no known functions in Alzheimer's disease. In this study, we used high-throughput small molecule screening in SH-SY5Y cells that stably expressed a luciferase reporter gene driven by the BACE1 (β-site amyloid precursor protein cleaving enzyme 1) promoter, which included a portion of the 5' untranslated region (5'UTR). We identified that CL82198, a selective inhibitor of MMP13, decreased BACE1 protein levels in cultured neuronal cells. This effect was dependent on PI3K (phosphatidylinositide 3-kinase) signalling, and was unrelated to BACE1 gene transcription and protein degradation. Further, we found that eukaryotic translation initiation factor 4B (eIF4B) played a key role, as the mutation of eIF4B at serine 422 (S422R) or deletion of the BACE1 5'UTR attenuated MMP13-mediated BACE1 regulation. In APPswe/PS1E9 mice, an animal model of Alzheimer's disease, hippocampal Mmp13 knockdown or intraperitoneal CL82198 administration reduced BACE1 protein levels and the related amyloid-β precursor protein processing, amyloid-β load and eIF4B phosphorylation, whereas spatial and associative learning and memory performances were improved. Collectively, MMP13 inhibition/CL82198 treatment exhibited therapeutic potential for Alzheimer's disease, via the translational regulation of BACE1.

Figure 1

CL82198 inhibits BACE1 protein expression through MMP13. (A) HEK293 cells were transfected with the luciferase reporter plasmids pGL4.21-BACE1 and pGL4.10 (negative control) in the absence or presence of 5 μM CL82198 or 10 nM WAY170523 for 48 h. Luciferase assays were performed with a GloMax 96 microplate luminometer. Firefly luciferase activity was normalized to that of the plasmid pGL4.10. (B) BACE1 was knocked down with BACE1 siRNA (siBACE-1 and -6) or control siRNA (NC) in HEK293 cells for 48 h. A dramatic decrease in the amount of BACE1 protein is shown at ∼70 kD. M = protein marker. (C) HEK293 cells were treated with WAY170523 (WAY, 10 nM) for 48 h, while control cells were treated with DMSO (CTRL). (D) SH-SY5Y cells were treated with 5 μM CL82198 for 6, 12, 24 and 48 h. (E) SH-SY5Y cells were treated with 1, 2.5, 5 and 10 μM of CL82198 for 48 h. (F) SH-SY5Y cells were treated with an MMP13 neutralizing antibody (anti-MMP13, 1:500) or control rabbit IgG antibody (CTRL) for 48 h. (G–I) Mmp13 was knocked down with shRNA-1 (shMMP13-1, G) or shRNA-2 (shMMP13-2, H) in HT22 cells for 72 h or was overexpressed with an MMP13 vector in HEK293 cells for 48 h (I). CTRL = control shRNA; MOCK = control vector. (J) HEK293 cells were transfected with either the control vector (MOCK) or MMP13 vector for 48 h in the absence or presence of 5 μM CL82198 (CL). Cell lysates were subjected to western blotting analysis. Representative western blots for BACE1 are shown on the top, and quantifications are shown below. (K) Primary cultured cortical neurons were treated with 0.1, 0.5, 1, 2, 5 and 10 μM SC205756 for 48 h. (L) HEK293 cells were treated with 1 μM SC205756 for 48 h. (M) HEK-APP cells were treated with 5 µM CL82198 (CL) for 48 h. Cell lysates were prepared and subjected to western blotting analysis for APP and ADAM10. sAPPβ was analysed in conditioned media using an sAPPβ antibody. All values were normalized to CTRL or MOCK (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA, n = 3 or 4).

Figure 2

MMP13-mediated BACE1 regulation involves PI3K signalling and is unrelated to BACE1 transcription and protein degradation. (A) BACE1 protein levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of the RTK inhibitor 341610 (4 μM). (B) SH-SY5Y cells were transfected with either control or MMP13 vector for 48 h in the absence or presence of 5 μM CL82198 (CL) and 4 μM 341610. (C) SH-SY5Y cells were transfected with either control or MMP13 vector for 48 h in the absence or presence of 4 μM 341610, and PI3K activity was measured by ELISA. (D) p-Akt protein in SH-SY5Y cells transfected with either control or MMP13 vector for 48 h in the absence or presence of 4 μM 341610. (E) BACE1 protein levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of the PI3K inhibitor LY294002 (LY, 5 and 10 μM). (F) SH-SY5Y cells were transfected with either control or MMP13 vector for 48 h in the absence or presence of 5 μM CL82198 (CL) and LY294002 (LY, 5 and 10 μM). (G) Relative BACE1 mRNA levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of LY294002 (LY, 5 and 10 μM). (H) Relative Bace1 mRNA levels in HT22 cells treated with vehicle (CTRL) or Mmp13 shRNA-1 for 72 h. (I and J) BACE1 protein levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of 1 μM MG132 (G) or 100 μM CQ (H). (K) BACE1 protein levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of the transcriptional inhibitor actinomycin D (ActD, 0.1 μM) or the protein synthesis inhibitor cycloheximide (CHX, 5 μM). All values were normalized to CTRL (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA, n = 3 or 4).

Figure 3

eIF4B phosphorylation at S422 is involved in MMP13-mediated BACE1 regulation. (A) BACE1 protein levels in SH-SY5Y cells treated with 5 μM CL82198 (CL) for 48 h in the absence (CTRL) or presence of 25 μM 4EGI1 (a competitive inhibitor of eIF4E/eIF4G). (B) Dose-response of phosphorylated eIF4B (S422, p-eIF4B) in SH-SY5Y cells treated with 0.1, 1, 2.5, 5 and 10 μM CL82198 for 48 h. (C) p-eIF4B levels in HEK293 cells transfected with either control or MMP13 vector for 48 h in the absence or presence of LY294002 (LY, 5 μM). (D) p-eIF4B levels in HEK293 cells transfected with shRNA control (CTRL) or Mmp13 shRNA-1 vector (shMMP13-1). (E) BACE1 proteins in HEK293 cells transfected with eIF4B S422R mutant (S422R) or wild-type eIF4B vector (eIF4B) in the absence or presence of 5 μM CL82198 (CL) for 48 h. (F) BACE1 protein levels in HEK293 cells transfected with control vector (MOCK), MMP13 or MMP13 and co-transfected with S422R or eIF4B. (G) BACE1 protein levels in HEK293 cells transfected with truncated human BACE1 with a 5′UTR deletion (−5′UTR) or with the entire human BACE1 sequence containing the 5′UTR (+5′UTR) in the absence or presence of 5 μM CL82198 (CL) for 48 h. (H) BACE1 protein levels in HEK293 cells expressing the −5′UTR (left two lanes) or +5′UTR construct (right two lanes) in which eIF4B or S422R was co-transfected. S422R decreased BACE1 protein levels in only the +5′UTR transfected cells. All values were normalized to CTRL or MOCK (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA, n = 3 or 4).

Figure 4

Mmp13 knockdown ameliorates Alzheimer’s disease-associated pathology in APP/PS1 mice. (A) Representative western blots and bar plot summary of MMP13 and PI3K expression in the prefrontal cortex of control (n = 13) and Alzheimer’s disease patients (n = 13). Data in (i and ii) and in (iii–v) were from samples provided by Sydney Brain Bank and NIH NeuroBiobank, respectively. (B) Immunohistochemical images and bar plot summary of neuritic plaques in the hippocampi of wild-type (WT) and APP/PS1 mice (AD) treated with vehicle (CTRL) and Mmp13 shRNA-1 (shMMP13-1). Neuritic plaques were probed with the amyloid-β-specific monoclonal antibody 6E10 (n = 6). (C) Representative western blots and bar plot summary of MMP13, APP, ADAM10, BACE1, PSEN1, βCTF, αCTF, sAPPβ and sAPPα expression in the hippocampi of wild-type (n = 8), wild-type with Mmp13 shRNA-1 (WT-shMMP13-1, n = 8), Alzheimer’s disease (n = 8) and Alzheimer’s disease with Mmp13 shRNA-1 mice (AD-shMMP13-1, n = 8). The soluble fractions were used to detect sAPPβ and sAPPα with sAPPβ and 6E10 antibodies. (D) Representative western blots and bar plot summary of MMP13 and BACE1 expression in the hippocampi of wild-type (n = 6), wild-type with Mmp13 shRNA-2 (WT-shMMP13-2, n = 6), Alzheimer’s disease (n = 6) and Alzheimer’s disease with Mmp13 shRNA-2 mice (AD-shMMP13-2, n = 6). (E) ELISAs were used to measure soluble and insoluble amyloid-β_40/42 levels in the brain homogenates of wild-type (n = 6), WT-shMMP13-1 (n = 6), Alzheimer’s disease (n = 6) and AD-shMMP13-1 (n = 6) mice. (F) Representative western blots and bar plot summary of p-eIF4B, eIF4B, neprilysin and insulin-degrading enzyme (IDE) levels in the hippocampi of wild-type (n = 8), WT-shMMP13-1 (n = 8), Alzheimer’s disease (n = 8) and AD-shMMP13-1 (n = 8) mice. All values were normalized to CTRL or wild-type (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA).

Figure 5

CL82198 ameliorates Alzheimer’s disease-associated pathology in APP/PS1 mice. (A) Immunohistochemical images and bar plot summary of neuritic plaques in the hippocampi of wild-type, wild-type with CL82198 (WT-CL), APP/PS1 (AD) and APP/PS1 mice treated with CL82198 (AD-CL) (n = 6). (B) Representative western blots and bar plot summary of MMP13, APP, ADAM10, BACE1, PSEN1, βCTF, αCTF, sAPPβ and sAPPα levels in the hippocampi of wild-type (n = 8), WT-CL (n = 8), AD (n = 8) and AD-CL (n = 8) mice. Soluble fractions were used to detect sAPPβ and sAPPα with sAPPβ and 6E10 antibodies. (C) ELISAs were used to measure soluble and insoluble amyloid-β_40/42 levels in the brain homogenates of wild-type (n = 6), WT-CL (n = 6), AD (n = 6) and AD-CL (n = 6) mice. (D) Representative western blots and bar plot summary of p-eIF4B, eIF4B, neprilysin and IDE levels in the hippocampi of wild-type (n = 8), WT-CL (n = 8), AD (n = 8) and AD-CL (n = 8) mice. All values were normalized to CTRL or wild-type (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA). Aβ = amyloid-β.

Figure 6

CL82198 treatment or Mmp13 knockdown improves learning and memory in APP/PS1 mice. Behavioural performance was assessed by Morris water maze tests followed by fear conditioning tests. (A) In the hidden platform tests, the time spent on reaching the platform (latency) was recorded by ANY-maze tracking software. Compared with vehicle-treated APP/PS1 mice (AD, n = 10), CL82198 treated APP/PS1 mice (AD-CL, n = 10) showed a significantly shorter latency on the third, fourth and fifth days. No significant differences were found for the AD-CL mice or wild-type mice treated with vehicle (WT, n = 8) or CL82198 (WT-CL, n = 9). (B and C) In the probe trial on the sixth day, AD-CL mice spent significantly more time travelling (B) and a significantly longer time staying (C) in the place where the hidden platform was previously placed than Alzheimer’s disease mice. (D and E) In the context fear conditioning tests, the freezing times and the total freezing times of the four groups of mice in the chamber were recorded and analysed after drug injection. (F) In the hidden platform tests, shMMP13-treated APP/PS1 mice (AD-shMMP13-1, n = 7) exhibited a significantly shorter latency on the third, fourth and fifth days than the vehicle-treated APP/PS1 mice (AD, n = 9). No significant differences were found between AD-shMMP13-1 and wild-type mice treated with vehicle (WT, n = 7) or shMMP13 (WT-shMMP13-1, n = 6). (G and H) On the sixth day of the probe trial, AD-shMMP13-1 mice spent significantly more time travelling (G) and a significantly longer time staying (H) in the place where the hidden platform was previously placed than the Alzheimer’s disease mice. (I and J) In the context fear conditioning tests, the freezing times and the total freezing times of the four groups of mice in the chamber were recorded and analysed for mice treated with shMMP13 or vehicle. All values were normalized to wild-type (1.0) within each experiment. The error bars are the SEM. n.s. = no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001 (ANOVA).

Figure 7

Schematic diagram depicting the possible mechanisms through which MMP13 regulates BACE1 translation. PI3K signalling downstream of MMP13 promotes eIF4B phosphorylation, which in turn facilitates the 5′UTR-dependent BACE1 translation. Consequently, increased BACE1 protein levels enhance amyloid-β production and impair cognitive function. The inhibition of MMP13 by CL82198 results in reduced BACE1 and amyloid-β accumulation in the brain, which contributes to the improved learning and memory functions in APP/PS1 mice. AD = Alzheimer;s disease; WT = wild-type.