3.1. Characterization of apoM-containing HDL particles in plasma
To determine the effects of ApoM in HDL on astrocyte cell death, we purified ApoM-containing HDL from plasma. After ultracentrifugation, the isolated HDL was purified on a HiTrap column with a monoclonal antibody against ApoM. Immunoblotting analysis showed that after 5 rounds of purification, ApoM could not be detected in HDL (Fig. 1A), indicating good preparations of ApoM-containing HDL and ApoM-free HDL. In addition, analysis using western blotting against ApoA-I, ApoB and ApoM demonstrated that HDL did not contain ApoB but contain rich ApoA-I (Fig.1B), consistent with the previous findings (Christoffersen, Nielsen et al. 2006). ApoM-containing HDL and ApoM-free HDL contained similar levels of ApoA-I, while HDL contain higher level of ApoA-I, suggesting the partial removal of ApoA-I during the purification processes.
3.2. Astrocytes undergo cell death following ischemic insult in vitro
When astrocytes were treated with deprivation of glucose and oxygen, cells underwent cell death (Huang, Zhang et al. 2008). As shown in Fig. 1C, the cell viability was reduced after 4-h OGD treatment, while 2-h OGD did not affect the viability [F (5, 12) = 85.205, P < 0.001], suggesting the time-dependent effects of OGD on cell death. To assess the cell viability after OGD and recovery, the cells were treated with 4-h OGD and recovery for 24 h, 48 h and 72 h. The results showed that 24-h recovery after 4-h OGD resulted in significant cell death [F (4, 35) = 66.317, P < 0.001; F (4, 35) = 118.808, P < 0.001 respectively], as evidenced by MTT reduction assay and LDH assay (Fig. 1D and 1E). As shown in Fig. 1F. after staining with hematoxylin, the astrocytes exposed to 4-h OGD and 24-h recovery exhibited cell injury. Then we choose 4-h OGD/24-h recovery as the treatment for astrocytes in the following study.
3.3. HDL with ApoM protects astrocytes against cell death induced by OGD/recovery.
To assess the role of ApoM-associated HDL in the OGD-induced astrocyte death, mouse HDL+ (ApoM-associated HDL) and HDL- (ApoM-depleted HDL) were isolated from C57 mouse plasma. After treated with OGD in the presence of HDL+ or HDL-, the cell viability was assessed by MTT assay. As shown in Fig. 2A and 2B, HDL+ at 10 ~ 20 mg/ml mitigated cell death and protected against LDH release induced by OGD, while no protective effects were observed for HDL- treatment [F(4, 35) = 99.296, P < 0.001; F(4, 35) = 108.535, P < 0.001; F(4, 35) = 183.831, P < 0.001; F(4, 35) = 90.188, P < 0.001 respectively]. In addition, the live cells were counted using trypan blue exclusion test after OGD treatment. As shown in Fig. 2C and 2D, OGD significantly decreased the number of living cells, which was reversed by HDL+ at 10 ~ 20 mg/ml [F(4, 25) = 42.651, P < 0.001]. Similarly, HDL- at 5 ~ 20 mg/ml did not show any effect [F(4, 35) = 26.627, P < 0.001].
3.4. ApoM-associated HDL restores the GFAP expression in astrocytes after OGD and recovery.
Astrocytes are generally more resilient than neurons after ischemic injury due to high level of antioxidant glutathione and less oxidative stress. (Almeida, Delgado-Esteban et al. 2002, Amri, Ghouili et al. 2017) In primary cultured astrocytes, OGD/recovery treatment or oxidative stress also induces significant cell death as well as expression change of GFAP, the specific marker of astrocytes (Chen and Liao 2002, Huang, Zhang et al. 2008, de Pablo, Nilsson et al. 2013), showing the disrupted astrocyte intermediate filament system.
As shown in Fig. 3A, using immunostaining with GFAP antibody, we found that the number of GFAP-positive astrocytes was remarkably decreased after 4 h-OGD and 24 h-recovery. Treatments with HDL+ at 10 ~ 20 mg/ml promoted cell survival and increased cell number, while HDL- did not exhibit any protective effects. In addition, immunoblotting analysis revealed that GFAP expression was reduced in the cells after OGD and recovery, which was restored by HDL+ but not HDL- (Fig.3B) [F(6, 35) = 58.014, P < 0.001].
3.5. ApoM associated HDL protects against astrocyte apoptosis after OGD/recovery.
During cerebral ischemia in mouse brain, astrocytes in the ischemic core undergo apoptosis and dysfunction resulted from oxidative stress, leading to the consequent neuronal death (Liu, Chen et al. 2014, Amri, Ghouili et al. 2017). In order to clarify the role ApoM in the protection by HDL against cell apoptosis, the astrocytes were stained with Hoechst 33258. The apoptotic astrocytes showed condensed nuclei and enhanced blue fluorescence. As shown in Fig. 4A and 4B, the astrocytes exhibited a weak apoptotic ratio but a significantly high apoptotic ration after OGD treatment. HDL+ decreased the percentage of apoptotic cells in a dose-dependent manner [F(6, 35) = 81.542, P < 0.001; F(4, 25) = 32.879, P < 0.001 respectively]. HDL+ at 10 ~ 20 mg/ml significantly protected against cell apoptosis compared to OGD only, while 10 ~20 mg/ml of HDL- treatment did not confer protection against cell apoptosis.
Under apoptosis, caspase 3 in cells will be cleaved, leading to the increased expression ratio of Bax and Bcl-2 (Petrache, Fijalkowska et al. 2006). As shown in Fig. 4C and 4D, OGD reduced the expression of Bcl-2 and induced the expression of Bax. Thus, the ratio of Bax/Bcl-2 was enhanced after OGD treatment, and administration of HDL+, not HDL-, reduced the expression ratio. In addition, the expression of cleaved caspase 3 (p17) was determined by immunoblotting in the astrocytes. The results showed that caspase 3 was cleaved after OGD treatment, showing the activation of caspase 3. Treatment with 10 mg/ml of HDL+ inhibited the cleavage of caspase 3, but HDL- did not show any inhibitory effects [F(4, 25) = 178.301, P < 0.001; F(4, 25) = 38.211, P < 0.001 respectively].
3.6. Akt and Erk signaling pathways participate in the anti-apoptotic effects of S1P in astrocytes challenged with OGD
It has been shown that PI3K/Akt and Erk signaling pathways are involved in the anti-apoptotic activities of S1P (Christoffersen, Obinata et al. 2011, Ruiz, Okada et al. 2017). To assess the role of Akt and Erk pathways in our model, we used LY249002 to inhibit PI3K/Akt and PD98059 to inhibit Erk activation. Those inhibitors were added to the astrocytes in the presence of HDL+ for 10 min (Akt activation) or 1 h (Erk activation). Then the astrocytes were lysed and the phosphorylation of Akt or Erk protein was analyzed by western blotting. As shown in Fig. 5A and 5B, HDL+ induced Akt phosphorylation which was abolished by LY 249002. Similarly, the upregulation of phosphorylated Erk by HDL+ was blocked by PD98059 [F(8, 45) = 124.488, P < 0.001; F(8, 45) = 101.441P < 0.001 respectively]. Thereafter, the astrocyte apoptosis after treated with LY249002 or PD98059 in the presence of HDL+ was determined by Hoechst 33258 staining. The results showed that blockage of Akt or Erk activation with specific inhibitors abolished the anti-apoptotic effect of HDL+ in astrocytes challenged with OGD (Fig. 5C and 5D) [F(3, 20) = 16.67, P < 0.001]. To confirm the results, we carried out Caspase-3 activity assay and found that HDL+ decreased Caspase 3 activity in the cells treated with OGD. However, LY249002 or PD98509 abrogated the protection of HDL+ (Fig. 5E) [F(3, 20) = 28.592, P < 0.001].
It has been reported that ApoM functions as a carrier of S1P in HDL, leading to the activation of downstream signaling of S1P after activating its receptors (Liu, Seo et al. 2014, Blaho, Galvani et al. 2015). In order to directly compare the effects of free S1P and ApoM-bound S1P, we loaded soluble mouse ApoM with S1P to obtain ApoM-bound S1P. Thereafter, we tested the Akt and Erk phosphorylations in the cells after treated with free S1P or ApoM-bound S1P. We found that ApoM-bound S1P induced expression of phosphorylated Akt or ERK comparable to that of free S1P, suggesting the similar effects of S1P bound or unbound to ApoM. However, we did not observe any effects of ApoM on cell apoptosis or Akt activation (data not shown).
3.7. S1pr1 but not S1pr3 is induced in astrocytes after OGD treatment
Upon activation, S1P exerts its biological effects through a group of G-protein coupled receptors, including S1P receptors (S1PR) 1-5. Thus, to determine the participation of S1P receptors in the anti-apoptotic effects of HDL+ on astrocytes, the expression of S1prs in astrocytes after OGD were assessed. It has been shown that S1pr1 and S1pr3 are main S1P receptors expressed in astrocytes (Pebay, Toutant et al. 2001), so only these two receptors were analyzed in this study.
After the cells were treated with OGD and recovery, the mRNA levels of S1pr1 and S1pr3 were assessed by real-time quantitative RT-PCR. The data showed that the mRNA expression of S1pr1 in the astrocytes was up-regulated after OGD treatment, while the S1pr3 level was not changed after OGD (Fig. 6A) [t(2) = 5.426, P < 0.001; t(2) = 10, P = 0.771 respectively]. To confirm the results, the protein levels of S1pr1 and S1pr3 were determined by immunoblotting. As shown in Fig. 6B, only S1pr1 protein expression was markedly induced in the astrocytes after OGD treatment [t(2) = 7.531, P < 0.001; t(2) = 10, P = 0.926]. These data suggest that S1pr1 may be required for the anti-apoptotic property of HDL+.
3.8. Activation of S1pr1 mediates protective effects of ApoM-associated HDL
Next, we determined which S1P subtype receptor is responsible for HDL anti-apoptotic property. Thus, we employed pharmacological antagonists of S1P receptors in the study. In the presence of HDL+, a S1pr1 antagonist W146 at 1 mM or a S1pr3 antagonist CAY10444 at 10 mM was added to the culture medium 30 min before OGD and kept in the medium until the end of the experiment. The Hoechst 33258 staining revealed that W146 abolished the protective effect of HDL+, while CAY 10444 did not show any effect (Fig. 7A) [F(2, 15) = 17.764, P < 0.001]. Likewise, as show in Fig 7B, the reduced Caspase 3 activity after treatment with HDL+ was inhibited by S1pr1 inhibition but not by S1pr3 inhibition [F(3, 20) = 25.754, P < 0.001]. Furthermore, immunoblotting analysis demonstrated that W146 inhibited the upregulation of phosphorylated Akt or Erk1/2, showing the abolishment of Akt/Erk1/2 signaling activation (Fig. 7C) [F(2, 15) = 121.98, P < 0.001; F(2, 15) = 114.062, P < 0.001 respectively]. These results suggest that S1pr1 but not S1pr3 activation is required for the anti-apoptotic effect of HDL+, which involves the activation of Akt/Erk1/2 signaling pathways.
3.9. Knockdown of S1pr1 abolishes the protection of ApoM-associated HDL
To exclude the possibility of unspecific effects of S1pr inhibitors, specific siRNAs were added to the astrocytes to silence the expression of S1pr1 and S1pr3. After the cells were treated with siRNAs for 3 days, the cell protein extracts were used for analysis of S1pr1 and S1pr3 by immunoblotting. As shown in Fig. 8A, immunoblotting results showed that protein expression of S1pr1 or S1pr3 was reduced by their respective siRNA. Thereafter, the apoptotic astrocytes after treatment with OGD were determined by Hoechst staining. We found that only S1pr1 silencing mitigated the anti-apoptotic effect of HDL+ in astrocytes (Fig. 8B) [F(3, 20) = 21.869, P < 0.001]. In addition, the Akt and Erk activation were determined by immunoblotting analysis. As shown in Fig. 8C, HDL+ or ApoM-bound S1P induced Akt and Erk phosphorylation, while the upregulation of phosphorylated Akt or Erk were blocked after S1pr1 silencing.