Isabelle Limon, UR4, Vieillissement, Stress et Inflammation, Université Paris 6, 7 quai St-Bernard, Bat A 5eme étage, 75252 Paris cedex 5, France. Tel.: +33 144 273 716; fax: +33 144 274 140; e-mail: email@example.com
Several studies have shown that the accumulation of β-amyloid peptides in the brain parenchyma or vessel wall generates an inflammatory environment. Some even suggest that there is a cause-and-effect relationship between inflammation and the development of Alzheimer's disease and/or cerebral amyloid angiopathy (CAA). Here, we studied the ability of wild-type Aβ1-40-peptide (the main amyloid peptide that accumulates in the vessel wall in sporadic forms of CAA) to modulate the phenotypic transition of vascular smooth muscle cells (VSMCs) toward an inflammatory/de-differentiated state. We found that Aβ1-40-peptide alone neither induces an inflammatory response, nor decreases the expression of contractile markers; however, the inflammatory response of VSMCs exposed to Aβ1-40-peptide prior to the addition of the pro-inflammatory cytokine IL-1β is greatly intensified compared with IL-1β-treated VSMCs previously un-exposed to Aβ1-40-peptide. Similar conclusions could be drawn when tracking the decline of contractile markers. Furthermore, we found that the mechanism of this potentiation highly depends on an Aβ1-40 preactivation of the PI3Kinase and possibly NFκB pathway; indeed, blocking the activation of these pathways during Aβ1-40-peptide treatment completely suppressed the observed potentiation. Finally, strengthening the possible in vivo relevance of our findings, we evidenced that endothelial cells exposed to Aβ1-40-peptide generate an inflammatory context and have similar effects than the ones described with IL-1β. These results reinforce the idea that intraparietal amyloid deposits triggering adhesion molecules in endothelial cells, contribute to the transition of VSMCs to an inflammatory/de-differentiated phenotype. Therefore, we suggest that acute inflammatory episodes may increase vascular alterations and contribute to the ontogenesis of CAA.
From a histo-pathological point of view, Alzheimer's disease (AD) is characterized by an age-dependent formation of amyloid β (Aβ)-containing plaques, accumulation of Tau protein-related neuro-fibrillary tangles (NFT), and neuronal loss in selective brain regions. Post-mortem brains of patients with AD as well as transgenic mouse models also display an increased expression of inflammatory mediators and, although much debated, several studies link the use of anti-inflammatory drugs with a reduction in risks of the disorder. Indeed, it has been shown that (i) there is a significant reduction in the activation of microglial cells in the brain of long-term NSAIDs users (Mackenzie & Munoz, 1998); (ii) disruption of memory by Aβ appears dependent on Cyclooxygenase-2 (COX-2)-mediated PGE2 signaling at the synapses, which is blocked by NSAIDs (Kotilinek et al., 2008); (iii) COX-2 expression is induced in neurons by Aβ, glutamate and inflammatory cytokines (Bazan, 2001; Blanco et al., 2010) and PGE2 levels are increased in patients with AD (Montine et al., 1999); and (iv) blocking interleukin-1β (IL-1β) signaling rescues cognition, attenuates Tau pathology and restores neuronal β-Catenin pathway function in an AD model (Kitazawa et al., 2011). Regarding this evidence, inflammation has been proposed either as ‘a driving force of Alzheimer disease’ or at least as ‘the third important component of the disease.’
Consistent with the importance of inflammation in AD, it has been shown that Aβ peptide-activated microglial cells trigger the release of inflammatory molecules directly toxic to neurons. Microglia-derived factors shown to be toxic to neurons include nitric oxide combined to superoxide anions, reactive oxygen species, tumor necrosis factor-α (TNFα), IL-1β, a protease-resistant toxin induced by interaction with microglial heparan sulfate, complement proteins, and cathepsin B. Moreover, activated microglia cells recruiting astrocytes actively enhance the inflammatory response triggered by extracellular Aβ deposits; they also initiate local cytokine-mediated acute-phase response activation of the complement cascade and induction of inflammatory enzyme systems such as the inducible nitric oxide synthase, multiple forms of PLA2, and the prostanoïd generated by COX-2; these factors, either alone or in concert, can contribute to neuronal dysfunction, microglial toxicity and cell death. Finally, many cytokines such as IL-1β, TNFα alone or combined with IFNγ, and chemokine signaling (CXCR2 signaling) promote Aβ production by modulating γ-secretase activity in neurons or increasing levels of endogenous BACE1 [β-Site APP cleaving enzyme-1, (Zhao et al., 2011)]. For review, the study described by Heneka et al., (2010) is referred.
Abnormal accumulation of Aβ not only occurs in brain parenchyma but also occurs in cerebrovasculature, either around the capillaries perfusing the cerebellum, cerebral cortex and leptomeninges [capillary cerebral amyloid angiopathy (CAA)], or within the media of medium and large arteries (arterial CAA) irrigating the central nervous system (Thal et al., 2008). Regarding the one form affecting the medial layer of the vessel wall, it is associated with vascular smooth muscle cells (VSMCs) degeneration, resulting primarily in a loss of tonus, and ultimately, in a disruption of the vessel wall integrity possibly leading to intracerebral hemorrhages and hypo-infusion (Attems et al., 2011). In the late 1990s, it had been suggested that degeneration of the cerebrovasculature resulting in hemorrhages may be at least partially mediated by vasculature inflammation. This hypothesis was built upon several observations. Monocyte/macrophage marker-positive foci/cells co-localized with HCHWA-Dutch arterial Aβ (Maat-Schieman et al., 1997); in sporadic CAA, cerebrovascular amyloid deposition, mainly composed of Aβ1-40 peptides, was associated with increased recruitment or activation of monocyte/macrophage lineage cells (Yamada et al., 1996); more recently, data in both mouse models and Aβ-related angiitis patients suggested a critical implication of cross-talk between endothelial cells, macrophages and T cells in the modulation of cerebrovascular amyloid deposition and CAA development (Weiss et al., 2011). In vitro, several Aβ-peptides induce both CD40 expression, IL-1β and interferon-γ secretion from human aortic endothelial cells; this latter finding also suggested that, although pro-inflammatory cytokines may be produced by recruited T cells and macrophages, Aβ also functions as a direct inflammatory stimulator of parietal cells (Suo et al., 1998). In addition, because cytokines modulate the expression of multiple genes including their own and those of their receptors, Suo et al.'s, data (1998) were the first suggesting that Aβ-induced cytokine production amplifies the initial Aβ-induced inflammation, being responsible for an amplification loop of the inflammatory molecular cascade. However, if the influence of Aβ peptides on endothelial cells has been well documented that of Aβ peptides on VSMCs inflammatory response remains ambiguous.
Here, we show that the inflammatory response of IL-1β-treated VSMCs was highly enhanced when previously exposed to wild-type Aβ1-40 peptide (Aβ1-40); Aβ1-40 peptide alone did not induce any inflammatory response in VSMCs. Using a pharmacological approach, we also demonstrate that this ‘sensitization’ process occurs through Aβ1-40 preactivation of the PI3kinase and possibly NFκB pathway and that it accelerates VSMCs de-differentiation. Finally, we evidenced that the inflammatory context generated by endothelial cells exposed to Aβ1-40 peptide has similar effects than the ones described with IL-1β, giving substance to a possible in vivo relevance to our findings.
Aβ1–40 does not induce any VSMCs inflammatory response
To determine whether Aβ1–40 accumulation could trigger an inflammatory response from VSMCs, we first assayed the expression or the secretion of two inflammatory markers, COX-2 and Prostaglandin E2. The concentration of Aβ1-40 peptide used was 50 μm and the length of treatment was 24 h. Treatment with IL-1β (10 ng mL−1), a pro-inflammatory cytokine, served as positive control and allowed VSMCs to fully trans-differentiate toward an inflammatory state (Clement et al., 2006). As shown in Fig. 1A,B, whereas IL-1β greatly induced the expressions of COX-2 and the secretion PGE2, Aβ1-40 was inefficient. Similar results were obtained when tracking MMP-9 (Fig. 1C,D), a matrix metalloprotease known to be involved in inflammatory diseases including peripheral arterial diseases (Busti et al., 2010). Indeed, MMP-9 mRNA levels (estimated by quantitative, RT–PCR Fig. 1C) and MMP-9 activity (measured by gelatin zymography, Fig. 1D) remained as undetectable in Aβ1-40-treated cells as in vehicle-treated cells (Ctl). As expected (Jung et al., 2003; Blaise et al., 2012), (i), MMP-2 mRNA levels was increased more than 6-fold (P <0.001) in cells treated with Aβ1-40, when compared to cells with vehicle (Fig. 1E); and (ii) the active-form of the MMP-2 protein (referred to as Act. MMP-2, Fig. 1D) was clearly detected in Aβ1-40-treated VSMCs. Of note, the bands observed in vehicle correspond to nonmature forms of the enzyme (Fig. 1D). Altogether, these results clearly evidenced that an Aβ1-40-treatment alone does not trigger VSMCs transition toward an inflammatory phenotype.
Aβ1–40 highly enhanced IL-1β-induced inflammatory response in VSMCs through PI3K and NFκB activation
Because Aβ-peptide accumulation generates vessel wall inflammation by inducing endothelial cell cytokine production and monocyte entry (Li et al., 2009; Vukic et al., 2009), we thought of revisiting the previous experiments mimicking this situation in vitro. This consisted of evaluating the inflammatory response of VSMCs in an amyloid context. To do so, VSMCs were sequentially treated with 50 μm of Aβ1-40 peptide and exposed to 2 ng mL−1 of IL-1β before measuring the expression and/or secretion of inflammatory markers. (Both treatments were 24 h; the details of the whole protocol are illustrated in Fig. 2A). As shown in Fig. 2B,C, 24-h-pretreatment with Aβ1-40 strongly enhanced the IL-1β-initiated-COX-2 expression and PGE2 secretion. Similar results were obtained monitoring MMP-9. Indeed, Aβ1-40-pretreatment also increased IL-1β-induced MMP-9 mRNA expression (Fig. 2D); this translated into potentiated MMP-9 enzymatic activity in VSMCs when compared to cells treated with IL-1β alone (over 4-fold, considering the MMP-9 bands normalized to pro-MMP2, Fig. 2E). To be mentioned, Aβ1-40 had a similar potentiating effect on MMP-9 and COX-2 expression when followed by TNFα treatment; nevertheless, the level of expression of COX-2 and MMP-9 in TNFα-treated VSMC is very weak and remains very low after the sensitization by Aβ1-40-treatment (Fig. S1). These results establish that the Aβ1-40 peptide sensitizes VSMCs to IL-1β and suggest that it may regulate one or several molecular entities involved in the IL-1β-induced inflammatory response of these cells.
Using pharmacological compounds such as the PI3K inhibitor LY294002, the selective (BAY11-7082) or nonselective (MG132) inhibitor of the NFκB pathway, we showed that the inhibition of either of these two pathways clearly abrogated the effect of IL-1β on COX-2 expression (Fig. 3A) and MMP-9 secretion (Fig. 3B). In addition, 50 μm of Aβ1-40 rapidly (within 15 min) triggered a significant phosphorylation of AKT reflecting a PI3K pathway activation, which decreased in less than 15 min (Fig. 3C). Finally, treating VSMCs transfected with the Igκ-CONA-Luc vector with Aβ1-40 for 24 h resulted in enhancing luciferase activity by ~ 12-fold when compared to vehicle-treated cells (P <0.001, Fig. 3D). As mentioned in the materials and methods section, the Igκ-Cona-Luc vector carries a luciferase reporter gene under the control of three synthetic copies of the NFκB consensus responsive element cloned upstream of the conalbumin transcription start site; the Cona-Luc control vector is identical, except that it does not contain NFκB sequences. The basal luciferase activity of the CONA-Luc control vector remained unchanged; as expected, a slight activity of the Igκ-CONA-Luc vector was detected in vehicle-treated cells, reflecting the basal activity of the NFκB pathway in VSMCs (Shin et al., 1996). Interesting enough, neither IκBα phosphorylation nor IκBα degradation, both usually characteristics of NFκB activation, was observed when Aβ1-40 treatment was performed for a short period of time (Fig. 3C). Altogether, these results (i) corroborate previous data obtained in SMC demonstrating that COX-2 and MMP-9 expressions can be induced by PI3K (Hsieh et al., 2006; Lee et al., 2007) and NFκB pathways (Duggan et al., 2007; Liang et al., 2007); and (ii) demonstrate that Aβ1-40 rapidly activates the PI3K pathway and belatedly triggers NFκB pathway activation. Therefore, we next evaluated whether Aβ1-40-activation of PI3K and NFκB pathways was involved in VSMCs sensitization to IL-1β.
To this end, VSMCs were first treated with 50 μm of Aβ1-40 for 24 h, with or without NFκB (MG132) or PI3K (LY294002) inhibitors; then, after removing the cell medium containing Aβ1-40 and inhibitors, VSMCs were exposed to 2 ng mL−1 of IL-1β for a 24-h-additional treatment before evaluating the expression of inflammatory markers (see Fig. 4A for the protocol). The reason for using only the MG132 proteasome inhibitor and not the specific NFκB inhibitor BAY11-7082 is that MG132 is a reversible compound, whereas BAY11-7082 is not; as PI3K and NFκB inhibitors abrogated IL-1β-induced expression of COX-2 and MMP-9 (Fig. 3A,B), the compounds used during Aβ1-40 pretreatment had to be reversible in order to avoid affecting IL-1β-activated pathways. As shown in Fig. 4B, inhibition of the PI3K pathway during Aβ1-40 cell exposure significantly reduced IL-1β-induced MMP-9 mRNA expression (P <0.001) (upper panel), MMP-9 secretion (middle panel), and COX-2 expression (lower panel) to levels observed for cells not subjected to Aβ1-40 before IL-1β treatment. Importantly, since attesting for the reversibility of LY294002, preincubation of the cells with this compound alone did not prevent the effect of IL-1β on MMP9-transcript (Fig. 4B upper panel, bar 6 vs. 2). Analogous results were obtained when performing similar experiments with the MG132 molecule (Fig. 4C). Lastly, we also observed that LY294002 and MG132, combined with Aβ1-40-pretreatment, annealed the Aβ1-40-dependent potentiation of PGE2 secretion initiated by IL-1β (Fig. 4D, bars 8 and 12 vs. 4, P <0.001). As a whole, this last set of experiments demonstrated that the increase in the IL-1β-dependent inflammatory response induced by Aβ1-40 exposure involves PI3K and NFκB activation.
Aβ1–40 highly enhanced IL-1β-induced VSMCs de-differentiation through PI3K and NFκB activation
Because vessel wall inflammation drives a SMC de-differentiation characterized by a loss of the contractile apparatus, we thought of examining this process in an amyloid context. To do so, cells were exposed to Aβ1–40 peptide for 48 h before being treated with IL-1β for 72 h; the loss of the contractile phenotype was tracked by measuring the expression of α-actin and SM22, two markers of differentiated contractile VSMCs (Shanahan et al., 1993). As expected, the expressions of α-actin and SM22 transcripts were decreased in cells treated with IL-1β alone (Fig. 5A, bar 2 vs. 1, P <0.001); however, this decrease was much more important when pre-exposed to Aβ1–40 peptide prior to IL-1β treatment (Fig. 5A, bar 4 vs. 2, P <0.001). Similar results were obtained when tracking protein expressions. Indeed, the intensities of α-actin and SM22 bands in Fig. 5B lane 4, (corresponding to proteins extracted from cells exposed to both Aβ1–40 peptide and IL-1β) were much lower than in Fig. 5B lane 2 (corresponding to proteins extracted from cells exposed to IL-1β alone); α-actin and SM22 labeling, as well as the number of α-actin and SM22-positive cells, were much lower in cells preincubated with Aβ1–40 prior to IL-1β (Fig. 5C, column 4) compared with cells exposed to IL-1β alone (Fig. 5C, column 2). Of note, similar results were obtained for SM-Calponin and SM-MHC (Fig. S2). Here again, the inhibition of PI3K pathway by LY294002 blocked the effect of Aβ1–40 on SM22 and α-actin transcript expressions in IL-1β-treated VSMCs. Indeed, the increased inhibition of α-actin and SM22 expressions returned to a level comparable to cells treated with IL-1β alone (Fig. 5A, bar 8 vs. 7, P <0.001). Consistently, LY294002 treatment translated into a marked recovery of the expression of both of these contractile proteins. Indeed, their corresponding bands after LY294002 treatment were almost as intense as in control/untreated cells (Fig. 5B, lane 8 vs. 1); analogous conclusions could be made out of immuno-cytochemistry experiments (Fig. 5C, column 8 vs. 1). Altogether these results demonstrated that Aβ1–40 exposure to VSMCs aggravates IL-1β's effect on VSMCs de-differentiation through a PI3K pathway preactivation.
Conditioned medium obtained from Aβ1–40-exposed endothelial cells culture increases the expression of inflammatory markers in Aβ1–40-exposed VSMCs and aggravates their de-differentiation
Previous studies have demonstrated that amyloid peptides-exposed endothelial cells produce pro-inflammatory mediators including cytokines (Suo et al., 1998; Vukic et al., 2009). Considering the vessel structure where endothelial cells dialog with VSMCs, we next investigated how the inflammatory context generated by endothelial cells (EC) exposed to Aβ1–40 peptide could influence Aβ1–40-enhanced VSMCs transition toward an inflammatory phenotype. Of note, ‘Amyloid’ VSMCs or ECs refer to VSMCs or ECs pre-exposed 48 h to Aβ1-40 peptide-; nonamyloid cells refer to VSMCs or ECs pre-exposed 48 h to Aβ1-40 peptide vehicle. To do so, we generated in vitro conditioned media by incubating mouse brain endothelial cells with or without 50 μm of Aβ1–40 peptide for 72 h; these condition media were referred to as CM-EC-Aβ or CM-EC-Ctl, respectively. As shown in Fig. 6B, ‘amyloid’ VSMCs were much more responsive to CM-EC-Aβ when compared to ‘nonamyloid’ VSMCs or ‘amyloid’ VSMCs receiving CM-EC-Ctl. Indeed, (i) whether at the messenger or at the protein level, the induction of COX-2 expression by CM-EC-Aβ was always significantly higher when VSMCs were previously exposed to Aβ1–40 (Fig 6B, bar & lane 4 vs. 2, P <0.01); (ii) similar observations could be made for PGE2 secretion (Fig 6C, bar & lane 4 vs. 2, P <0.01); and (iii) although CM-EC-Ctl triggered an increase in COX-2 expression and PGE2 secretion from ‘amyloid’ VSMCs (the reference being the effect of CM-EC-Ctl on vehicle-treated VSMCs, Western blot Fig. 6B lane 3 vs. 1 and Fig. 6C, bar 3 vs. 1, P <0.01), CM-EC-Aβ systematically have a significantly higher effect on these parameters when compared to CM-EC-Ctl (Fig. 6B and C, bar & lane 4 vs. 3, P <0.01).
When tracking VSMCs de-differentiation (by measuring SM22 and α-actin transcript expressions Fig. 6D left and right panels, respectively), we also evidenced that Aβ1–40 exposition prior to CM-EC-Aβ aggravates the deleterious effects of inflammation (bars 4 vs. 3, P ≤0.05). Of note, CM-EC-Aβ added to ‘nonamyloid’ VSMCs did not have any effect when compared to CM-EC-ctl (Fig. 6D, bar 2 vs. 1); conversely, CM-EC-ctl on ‘amyloid’ VSMCs triggered a significant decrease in these markers (Fig. 6D, bar 3 vs. 1, P <0.001).
When revisiting these experiments in presence of LY294002, we further emphasized the importance of the PI3K pathway in the Aβ1–40-sensitization process of VSMCs to an inflammatory context. Indeed, here again LY294002 annealed the CM-EC-Aβ induced raised of COX-2 expression (Fig 6.B bar & lane 8 vs. 4, P <0.01) and PGE2 secretion (Fig. 6C bar 8 vs. 4, P <0.001). Interestingly enough, the small but significant rise of COX-2 expression and PGE2 secretion induced by CM-EC-ctl on Aβ1–40-treated VSMCs is also completely blocked by incubation with the LY294002 compound (Fig. 6B Western blot lane 7 vs. 3 and Fig. 6C bar 7 vs. 3, P <0.01). Finally, PI3K inhibition also prevented ‘amyloid’ VSMCs sensitization to the inflammatory context generated by EC exposure to Aβ1–40; indeed, it restored α-actin and SM22 transcript expressions to levels similar to vehicle-treated VSMCs receiving CM-EC-Aβ (Fig. 6D bar 8 vs. 4, P <0.01 and P ≤0.05). Altogether, these results (i) corroborate published data showing that endothelial cells exposed to an amyloid context generate on their own a strong inflammatory response (Suo et al., 1998; Vukic et al., 2009); (ii) reinforce the hypothesis as to whether ‘amyloid’ VSMCs, being at the vicinity of endothelial cells, amplify the initial Aβ-induced inflammation, therefore possibly responsible for an amplification loop of the vascular inflammatory statute; and (iii) underline the importance of PI3K pathway in these processes.
It has been shown that CAA-related inflammation might contribute to the pathogenesis of the disease. However, the effects of amyloid deposition on VSMCs inflammation and the molecular mechanisms involved remain unknown. Here, we first demonstrated that the Aβ1-40 peptide does not induce an inflammatory response in VSMCs because Aβ1-40 peptide treatment did not result in PGE2 or MMP-9 secretion or in COX-2 expression. However, Aβ1-40-peptide treatment clearly hypersensitized VSMCs to a pro-inflammatory context. Indeed, when exposed to the pro-inflammatory cytokine IL-1β οr a medium conditioned by ‘amyloid’ endothelial cells, ‘amyloid’ VSMCs over-expressed and/or released these above-mentioned parameters; moreover, this phenomenon was accompanied with a decreased expression of both smooth muscle α-actin and SM22, two contractile markers. Therefore, we now claim that inflammation in vessels displaying Aβ1-40 amyloid deposits, although beginning with endothelial cells is amplified through ‘amyloid’ VSMCs. This hypothesis is further supported by the fact that, in our model, this amplification only starts when Aβ1-40 concentration reach 25 μm, corresponding to the appearance of oligomeric forms (Fig. S3). This phenomenon, associated with VSMCs de-differentiation ultimately leading to VSMCs degeneration could cause, or at least aggravate, the vascular alterations characteristic of CAA, resulting in hemorrhage, cerebral hypo-perfusion, and stroke. It could also play a fundamental role in the development of Alzheimer's disease and be responsible for associated neuro-degeneration and cognitive impairment.
The lack of effect of Aβ1-40 alone on VSMC-contractile proteins expression objectified here (Fig. 5) is consistent with Chow et al.'s results; indeed, although using Aβ1-42, they showed in cultured VSMC that exogenous Aβ-peptide did not affect the expression of the serum responsive factor (SRF), an interacting transcription factor that orchestrate a VSMC-contractile/differentiated phenotype (Chow et al., 2007). This further support the hypothesis as to whether the increase in SRF observed in AD VSMC is not a consequence of accumulated Aβ-peptides.
The absence of inflammatory response from Aβ1-40-treated VSMCs is consistent with Suo et al.'s data showing that, among parietal cells, Aβ1-40 peptide mainly triggered the production of pro-inflammatory mediators from endothelial cells (Suo et al., 1998). However, Previti et al. (2006) reported that VSMCs exposed to the E22Q Dutch–mutant Aβ1-40 (Dutch Aβ1-40)-peptide secrete significant amounts of interleukin-6 (IL-6) (Previti et al., 2006). Similar to IL-1β, members of the IL-6 cytokine family including IL-6, IL-11, and oncostatin M, increase the expression of COX-2 and promote the release of PGE2 in several cellular models (Tai et al., 1997; Osuka et al., 1998; Bernard et al., 1999); in addition, they are thought to mediate the effect of IL-1β on COX-2 expression in SMCs (Lahiri et al., 2001). Rather than being controversial, we believe that these data support, once again, a possible difference between the ontogenesis of Dutch familial CAA and sporadic forms of CAA (characterized by Dutch and WT-Aβ1–40 deposition, respectively) which we recently reported (Blaise et al., 2012).
The molecular mechanism responsible for the hyper-sensitization of ‘amyloid’ VSMCs involves Aβ1-40 preactivation of signaling pathways triggering COX-2 and MMP-9 expression in VSMCs, namely the PI3Kinase and possibly NFκB pathways (Hsieh et al., 2006; Duggan et al., 2007; Lee et al., 2007; Liang et al., 2007). Indeed, Aβ1-40 treatment of VSMCs induces AKT phosphorylation and NFκB activity (Fig. 3) and their inhibition (conducted with LY294002 and MG132, respectively) abrogates this phenomenon (Fig. 4). Whether the MG132 effect could be attributed to the inhibition of NFκB pathway could be a matter of debate as this compound inhibits the 26S complex of the proteasome, thus affecting the degradation of numerous ubiquitinated proteins and, consequently, many cellular processes [see (Demasi & Laurindo, 2012) for review]. Nevertheless, if Aβ1–40 peptide actually induces NFκB signaling, this effect would likely depend on PI3kinase activation. Indeed, (i) direct activation of NFκB usually occurs within 15 min in VSMCs (Katsuyama et al., 1998) and this is not the case here (see Fig. 3C); (ii) the Aβ1–40-induced NFκB activation occurs later than Aβ1–40-induced PI3Kinase activation (Fig. 3C); and (iii) several studies have demonstrated a cross-regulation between PI3K and NFκB pathways (Reddy et al., 2000; Madrid et al., 2001). Even more convincing, Cheng et al. (2010) showed that the inhibition of PI3K by LY294002 compound attenuates the IL-1β-induced recruitment of the activated p65 subunit of NFκB to the MMP-9 promoter region (Cheng et al., 2010).
Because PI3K/Akt pathway is involved in the induction of IL-1-Receptor1 (Teshima et al., 2004), we proposed that one of the mechanisms involved in the hypersensitization of ‘amyloid’ VSMCs to IL-1β includes the increase in expression of this receptor. This scenario is strengthened by our results showing that Aβ1-40-treatment significantly enhanced the expression of IL-1-receptor1 (Fig. S4). Of note, the hypothesis as to whether Aβ1-40 exposition enhances the expression of receptors binding pro-inflammatory mediators was already enounced by Suo et al. (1998) and served to explain how Aβ1-40-treated VSMCs may contribute to the development of an inflammatory process within vessels.
Considering that Aβ1–40 alone significantly activates the PI3kinase/Akt and possibly NFκB pathways, the lack of its effect on MMP-9 and COX-2 expressions could be surprising (Fig. 1 and Fig. 3). In fact, this lack is consistent with the need of being triggered by an inflammatory context; it also reveals that IL-1β induces distinct signaling pathways. Because p42/p44 MAPK, p38 MAPK, and the JNK pathways have been shown to be involved in IL-1β-induced COX-2 and MMP-9 (Bartlett et al., 1999; Laporte et al., 1999; Liang et al., 2007) expression, we hypothesize that they are the pathways activated here by IL-1β.
Adding medium conditioned by ‘amyloid’ endothelial cells to Aβ1-40-exposed VSMCs potentiates the expression of inflammatory markers and further down-regulates contractile markers. The fact that ‘amyloid’ VSMCs were hypersensitized to a medium conditioned by Aβ1-40-exposed endothelial cells highlights the possible in vivo relevance of our in vitro study. More accurately, it also substantiates that endothelial cells exposed to Aβ-peptides release pro-inflammatory mediators (Suo et al., 1998; Vukic et al., 2009). Interestingly, conditioned medium derived from Aβ1-40-treated endothelial cells did not have any effect on control (i.e Aβ1-40 unexposed) VSMCs; this emphasizes the possible importance of the amyloid content within the vessel in increasing the inflammation due to VSMCs. In this regard, it would be much of interest to determine how a decrease in LRP-mediated amyloid-β clearance influences the inflammation-related CAA progression; AD individuals with CAA showed elevated levels of SRF and MYOCD, Aβ accumulation and significantly lower levels of LRP, compared with age-matched, nondemented controls without CAA (Bell et al., 2009).
As the effect of conditioned medium derived from Aβ1-40-treated endothelial cells is very similar the effect of IL-1β on ‘amyloid’ VSMCs, one may suggest that IL-1β is a major component of this medium. Nevertheless, the pro-inflammatory cytokines IL-1β, IL-2, IL-6, TNFα and IFN-γ were poorly detected whether in vehicle or in Aβ1-40-treated endothelial cell conditioned medium (Fig. S5). On the other hand, we evidenced a reproducible increased secretion of LIX (CXCL5) and MCP-1 from Aβ1-40-treated endothelial cells compared to vehicle-treated endothelial cells (fold 1.5 ± 0.035 and 1.44 ± 0.66, respectively). Because CXCL5 is a small cytokine produced in response to IL-1β (Chang et al., 1994), one may suggest CXCL5 as the relaying cytokine responsible for ‘amyloid’ VSMCs phenotypic transition toward a de-differentiated/inflammatory state. In partial support of this suggestion, one of its receptors (namely CXCR2) is involved in airway SMC trans-differentiation (Al-Alwan et al., 2012). It is also likely that MCP-1 participates to the phenotypic modification of ‘amyloid’ VSMCs; indeed, it is directly involved in inducing a phenotypic transition of VSMCs through the PI3K pathway (Selzman et al., 2002).
In conclusion, these in vitro experiments combined with the use of medium conditioned by endothelial cells on VSMCs allow a close approximation of the in vivo situation driving inflammation in CAA; however, macrophages and cell systems are also major players in the inflammatory process. Therefore, experiments need to be conducted on transgenic mouse models which develop cerebral and vascular amyloid plaques [APP23, (Winkler et al., 2001)] to accurately study the vascular changes and the inflammatory processes taking place during CAA. In addition, the importance of MCP-1 and LIX secretion by ECs exposed to Aβ1-40 peptide should be evaluated by performing experiments similar to those described in this study by either inhibiting MCP-1 or LIX expression in endothelial cells, or by blocking their receptors in VSMCs.
The source of reagents is given in Table S1 (Supporting information). The Aβ1–40 peptide purity is > 95%. Aβ1–40 peptide was solubilized at 2 mm in ultrapure water and frozen. The same batch was used for all experiments.
Mouse brain endothelial cells (bEnd.3/LGC Standards, Molsheim, France) were grown in Dulbecco‘s Modified Eagle‘s Medium (4.5 g L−1 glucose), 10% FCS and antibiotics. Rat VSMCs were isolated as described by Blaise et al., (2012). Experiments were performed on cells at passages from 2 to 6 for VSMCs or from 3 to 13 for bEnd.3 cells. Confluent cells were made quiescent growing them in a serum-free medium 12 h before treatment. Experiments were performed in serum-free medium. Media conditioned by endothelial cells were centrifuged 5 min 10 000 × g before contacting them with VSMCs. Conditions of cell treatment are indicated in the Figure legends. All incubations were performed at 37 °C in a 5% CO2 atmosphere. Cells were exposed to Aβ1–40-peptide diluted in culture medium at 50 μm for 24–48 h.
Total RNA was extracted from VSMCs using the RNeasy kit (Qiagen, Hilden, Germany). RT-PCR assays were performed as described by Blaise et al., (2012);. The forward and reverse primer sequences used to amplify the cDNA are given in Table 1.
sc-79079; Santa-Cruz Biotechnology, Inc. (Tebu-Bio, Le Perray en Yvelines, France)
sc-1745; Santa-Cruz Biotechnology, Inc.
sc-8040; Santa-Cruz Biotechnology, Inc.
sc-871; Santa-Cruz Biotechnology, Inc.
sc-20357; Santa-Cruz Biotechnology, Inc.
Zymography, PGE2, and luciferase assays
Gelatin zymography of conditioned media samples was performed as described by Blaise et al., (2012). PGE2 secretion was evaluated with an enzyme immunoassay kit from Cayman Chemical SPI-BIO (Massy, France). For luciferase activity, VSMCs were transfected by adding a total DNA/Fugen HD ratio of 2:3 according to the Roche Diagnostics instructions. VSMCs were transfected with 1 μg of firefly-luciferase reporter plasmid-Igκ-cona- or control vector cona-luc- and 50 ng of pRL-TK (Renilia-luciferase reporter gene under the control of thymidine-kinase promoter, as internal control). Transfected cells were treated for 24 h with Aβ1-40 peptide. Cells were harvested in reporter lysis buffer; lysates were assayed for luciferase activities using a dual luciferase assay kit (Promega, Lyon, France) and were normalized by the ratio of firefly and Renilla luciferase activities. The firefly-luciferase reporter plasmid- Igκ-cona-luc made of 3 NFκB sites cloned upstream of the minimum conalbumin promoter comes from Dr R.Weil, Institut Pasteur, Paris.
Immunocytochemistry was performed as described by Blaise et al., (2012). After treatment, cells were washed, fixed, permeabilized and incubated with primary antibodies (anti-α-actin or -SM22). Incubation with DyLight 488- and DyLight 549-conjugated secondary antibodies was performed for 1 h, Hoescht staining for 5 min at RT. Coverslips were mounted with Dako mounting medium (Dako, Carpinteria, CA, USA). Cells were examined on a Leica SP5 confocal fluorescent microscope. Acquisition parameters were established for vehicle-treated cells and were unchanged for all experimental conditions.
Data are reported as the mean ± SD. Values were compared between groups with the Welch's unpaired, corrected t-test.
We thank C. Xydas for technical assistance. This project has been supported by ‘Pierre Fabre Innovation’.
I. Limon and R. Blaise have access to all data and take responsibility for data and accuracy of the analysis. All the authors approved the final manuscript version. R. Blaise, G. Béréziat and I. Limon contributed to the study conception and design. R.Blaise, C. Rouxel, N. Trabelsi and A. Vromman contributed to the acquisition of the data. R. Blaise, I. Limon and A. Vromman contributed to the analysis and interpretation of the data. R. Blaise and I. Limon contributed to the manuscript preparation. R. Blaise and A. Vromman contributed to the statistical analysis.