Tested Compounds, Radical Scavenging and Cytotoxicity
Coumarin derivatives LM-021, LMDS-1 and LMDS-2, and pharmacological chaperone tafamidis (Fig. 1a) were examined. Based on molecular weight (MW, 335.35 − 366.25, respectively), hydrogen bond donors (HBD, 1 − 0), hydrogen bond acceptors (HBA, 5 − 4), and calculated partition coefficient of octanol/water (cLogP, 5.00 − 2.73), these four compounds meet Lipinski’s criteria in predicting oral bioavailability (MW ≤ 450, HBD ≤ 5, HBA ≤ 10, cLogP ≤ 5) [24] (Fig. 1b). In accordance with calculated polar surface area (PSA) of 66.8 − 52.6 Å2 (Fig. 1b), all four compounds displayed potential of blood–brain barrier (BBB) penetration (< 90 Å2) [25]. With predicted BBB permeation score (0.027 − 0.161) greater than that of threshold (0.02) [26], all four compounds were predicted to be BBB permeable by online BBB predictor.
The free radical scavenging activity of the test compounds was examined using substrate 1,1-diphenyl-2-picrylhydrazyl (DPPH) and kaempferol as a positive control [27]. While LMDS-1, LMDS-2 and tafamidis displaying weak DPPH radical scavenging activity (EC50 values: 257 − 969 µM), kaempferol and LM-021 had EC50 values of 35 and 79 µM, respectively (Fig. 1c). In addition, oxygen radical antioxidant capacity was determined based on the trolox standard curve. At 4 − 100 µM concentration, LM-021, LMDS-1, LMDS-2 and tafamidis had trolox equivalent activity of 11 − 28, 4 − 13, 4 − 10 and 5 − 10 µM, respectively (Fig. 1d). The cytotoxicity of LM-021, LMDS-1, LMDS-2 and tafamidis in mouse BV-2 microglia and human SH-SY5Y cells was examined by using MTT assay. Figure 1e showed that all four test compounds demonstrated cell viability of 73–70% and 94–88%, respectively in 100 µM compound-treated BV-2 and SH-SY5Y cells.
Anti-Inflammatory Potentials of Test Compounds in LPS/IFN-γ Induced Microglial Activation
Activated microglia and pro-inflammatory cytokine IL-1β have been found in pons of SCA3 patients [10]. In addition, GAL3 is involved in cellular inflammation mediated by NLRP3 inflammasome [28]. Therefore, we examined the anti-inflammatory potentials of test compounds using LPS/IFN-γ-stimulated mouse BV-2 microglia (Fig. 2a). LPS/IFN-γ treatment activated BV-2 cells, as revealed by the transformed morphology (Fig. 2b) along with increased release of NO (from 0.8 µM to 20.8 µM), IL-1β (from 5.8 pg/mL to 12.5 pg/mL), IL-6 (from 5.8 µg/mL to 15.6 µg/mL) and TNF-α (from 0.3 µg/mL to 3.8 µg/mL) in culture medium (p < 0.001) (Fig. 2c), as well as increased expression of NLRP3 (from 100–236%), GAL3 (from 100–154%), IBA1 (from 100–123%) (Fig. 2d), CD68 (from 100–138%) and MHCII (from 100–174%) (p = 0.002–<0.001) (Fig. 2e). Moreover, treatment with the test compounds (10 µM) reduced the levels of NO (from 20.8 µM to 16.7–15.2 µM; p < 0.001), IL-1β (from 12.5 pg/mL to 7.9 − 5.6 pg/mL; p = 0.01−<0.001), IL-6 (from 15.6 µg/mL to 12.1 − 8.7 µg/mL; p = 0.004−<0.001), TNF-α (from 3.8 µg/mL to 2.4 − 2.1 µg/mL; p < 0.001), NLRP3 (from 236% to 94 − 84%; p < 0.001), GAL3 (from 154% to 97 − 71%; p < 0.001), IBA1 (from 123% to 111 − 99%; p = 0.126 − 0.001), CD68 (from 138% to 115 − 113%; p = 0.017 − 0.01) and MHCII (from 174% to 147 − 132%; p = 0.048 − 0.002) (Fig. 2c − e).
Neuroprotective Effects of Test Compounds on ATXN3/Q75-GFP SH-SY5Y Cells
ATXN3/Q75-GFP SH-SY5Y cells [20] were used to examine the neuroprotective effects, including reduction of cellular oxidative stress and associated polyQ aggregation as well as promotion of neurite outgrowth, of test compounds (Fig. 3a). Expression of pathogenic ATXN3/Q75-GFP protein in retinoic acid-differentiated SH-SY5Y cells with or without test compound addition was insufficiently toxic to cause cell death (98–100% vs. 99%, p > 0.05) (Fig. 3b). However, the expressed ATXN3/Q75-GFP protein increased cellular ROS in SH-SY5Y cells (133%, p = 0.005), whereas application of LM-021, LMDS-1, LMDS-2 or tafamidis attenuated cellular ROS (102–97% vs. 133%, p = 0.005–0.002) (Fig. 3c). In ATXN3/Q75-GFP-expressing cells, treatment with LM-021, LMDS-1, LMDS-2 or tafamidis led to 15–25% reduction of aggregation (from 9.0% to 7.6–6.8%, p < 0.001) (Fig. 3d). In addition, increased neurite length (from 22.2 µm to 27.7–36.2 µm, p = 0.047–<0.001), process (from 2.3 to 2.8–3.1, p < 0.001) and branch (from 1.5 to 2.1–2.9, p = 0.017–<0.001) was observed in LM-021, LMDS-1, LMDS-2 or tafamidis-treated cells (Fig. 3e).
Inflammatory Priming of ATXN3/Q75-GFP SH-SY5Y Cells with BV-2 Conditioned Medium, MDP, IL-1β or IL-6
As shown in Fig. 2, LPS/IFN-γ treatment increased the amount of IL-1β (12.5 pg/mL) and IL-6 (15.6 µg/mL) in culture media of BV-2. To clarify the inflammatory priming role of IL-1β and IL-6 in BV-2 conditioned medium, IL-1β in 8 pg/mL or IL-6 in 10 µg/mL was added to ATXN3/Q75-GFP-expressing SH-SY5Y cells to mimic the concentration of IL-1β or IL-6 in BV-2 conditioned medium added to SH-SY5Y cells. MDP, a NLRP1 agonist upregulating Aβ-associated inflammation in the SH-SY5Y cells [29], was also included for comparison. After induction of ATXN3/Q75-GFP expression in SH-SY5Y cells for 6 days, the culture medium was removed and replaced with new cultural medium/conditioned medium (± LPS/IFN-γ stimulation, 2:1 ratio), MDP (0.1 µg/mL), IL-1β (8 pg/mL) or IL-6 (10 µg/mL), and further incubated for 48 h (Fig. 4a). Figure 4b shows that addition of CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP SH-SY5Y cells led to reduction of the cell viability compared to induction only (from 95–88%, p < 0.001), and the reduction was also significant compared to addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 (88% vs. 94–95%, p < 0.001). Moreover, addition of CM (+ LPS/IFN-γ) lead to increase of caspase 1 activity (from 100–831%, p < 0.001) and LDH release (from 100–1215%, p < 0.001) compared to induction only, and the increase was also significant compared to addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 (caspase-1: 831% vs. 336–264%; LDH release: 1215% vs. 664–348%) (p < 0.001). Compared to induction only, the cell viability was not reduced upon addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 (94–95% vs. 95%, p > 0.05), although caspase 1 activity (336–264% vs. 100%, p = 0.019–0.001) and LDH release (664–348% vs. 100%, p = 0.005–<0.001) did increase.
ATXN3-containing polyQ expansion may increase oxidative stress [30]. In ATXN3/Q75-expressing cells, addition of CM (+ LPS/IFN-γ) dramatically increased cellular ROS level (from 119–178%, p < 0.001) and ATXN3/Q75 aggregation (from 7.0–12.9%, p < 0.001), and the increase was also significant compared to addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 (ROS: 178% vs. 145–118%; ATXN3/Q75 aggregation: 12.9% vs. 8.6–7.4%) (p < 0.001). While MDP and IL-1β addition increased cellular ROS (145–141%, p = 0.006–0.002), change of aggregation upon addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 was not significant (8.6–7.4% vs. 7.0%, p > 0.05) (Fig. 4c, d). Lastly, addition of CM (+ LPS/IFN-γ) to ATXN3/Q75-expressing cells significantly reduced neurite length (from 25.7 µm to 16.7 µm), process (from 2.1 to 1.7) and branch (from 1.4 to 0.9) in comparison with the untreated cells (p < 0.001), and the reduction was notable or significant compared to addition of CM (-LPS/IFN-γ), MDP, IL-1β or IL-6 (length: 16.7 µm vs. 20.4–23.6 µm, p = 0.097–<0.001; process: 1.7 vs. 1.9–2.0, p = 0.037–0.007; branch: 0.9 vs. 1.1–1.3, p = 0.01–<0.001) (Fig. 4e). Treatment of MDP, IL-1β or IL-6 also reduced neurite length (21.4–20.0 µm vs. 25.7 µm, p = 0.02–0.003) and branch (1.1 vs. 1.4, p = 0.008–0.006). Based on these results, CM (+ LPS/IFN-γ) induced the strongest cytotoxicity effects and was therefore selected to inflame ATXN3/Q75-GFP-expressing cells for examining the anti-inflammatory of test compounds.
Effects of Test Compounds on CM (+ LPS/IFN-γ)-Inflamed ATXN3/Q75-GFP SH-SY5Y Cells
Before inducing ATXN3/Q75 expression for 6 days, retinoic acid-differentiated ATXN3/Q75-GFP SH-SY5Y cells were pretreated for 8 h with test compounds (10 µM). BV-2 CM (± LPS/IFN-γ) added to new cultural medium at a 2:1 ratio was applied to the cells on the sixth day for 2 days (Fig. 5a). Figure 5b shows that BV-2 CM (+ LPS/IFN-γ) addition reduced the viability of ATXN3/Q75-GFP SH-SY5Y cells (from 95–87%, p < 0.001), and application of LM-021, LMDS-1, LMDS-2 or tafamidis rescued the decreased cell viability (94% vs. 87%, p < 0.001). Addition of BV-2 CM (+ LPS/IFN-γ) also increased caspase 1 activity (from 100–489%, p < 0.001) and LDH release (from 100–202%, p < 0.001) of ATXN3/Q75-GFP SH-SY5Y cells, whereas application of the test compounds attenuated the caspase 1 activity (355–309% vs. 489%) and LDH release (120–107% vs. 202%) (p < 0.001). In addition, significantly increased ROS production (from 100–135%, p < 0.001) was observed in ATXN3/Q75-GFP SH-SY5Y cells added with BV-2 CM (+ LPS/IFN-γ), and all the four test compounds ameliorated oxidative stress induced by BV-2 CM (+ LPS/IFN-γ) addition (115–102% vs. 135%, p = 0.007–<0.001) (Fig. 5c). Addition of BV-2 CM (+ LPS/IFN-γ) also significantly increased ATXN3/Q75 aggregation compared to the untreated cells (from 9.2–13.9%, p < 0.001), and treatment of test compounds led to 23–30% reduction of aggregation (from 13.9% to 10.7–9.8%, p < 0.001) in inflamed ATXN3/Q75-GFP-expressing cells (Fig. 5d). Moreover, addition of BV-2 CM (+ LPS/IFN-γ) reduced neurite length (from 19.1 µm to 13.0 µm, p < 0.001), process (from 2.3 to 2.0, p = 0.005) and branch (from 1.6 to 1.0, p = 0.002) in ATXN3/Q75 cells compared to BV-2 CM (-LPS/IFN-γ)-treated cells, and treatment of test compounds increased neurite length (15.8–19.1 µm vs. 13.0 µm, p = 0.005–<0.001), process (2.2–2.3 vs. 2.0, p = 0.027–0.003) and branch (1.5–1.7 vs. 1.0, p = 0.011–<0.001) (Fig. 5e). These results demonstrate that test compounds could protect cells from cell death, reduce cellular ROS and ATXN3/Q75 aggregation, and improve neurite outgrowth in BV-2 CM (+ LPS/IFN-γ)-inflamed ATXN3/Q75-GFP-expressing cells.
Down-Regulation of NLRP1 Inflammasome Pathways by Test Compounds in BV-2 CM (+ LPS/IFN-γ)-Inflamed ATXN3/Q75-GFP SH-SY5Y Cells
In addition to NLRP3, NLRP1 inflammasome also promotes the IL-1β activation [31]. We thus examined the expression of NLRP1, ASC (PYD and CARD domain containing adaptor) and IL-1β in BV-2 CM (+ LPS/IFN-γ)-inflamed ATXN3/Q75-GFP SH-SY5Y cells by immunoblotting using specific antibodies. As shown in Fig. 6a, addition of BV-2 CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP-expressing SH-SY5Y cells raised the expressions of NLRP1 (from 116–147%, p = 0.092), ASC (from 129–162%, p = 0.507) and IL-1β (from 114–214%, p < 0.001), whereas LM-021, LMDS-1, LMDS-2 or tafamidis treatment reduced the levels of NLRP1 (106 − 90% vs. 147%, p = 0.017−<0.001), ASC (115 − 96% vs. 162%, p = 0.153 − 0.017) and IL-1β (158 − 120% vs. 214%, p = 0.01−<0.001).
Upon binding to its receptor, IL-1β is capable of activating JNK/JUN, NF-κB (P65/P50 heterodimer) and P38/STAT1 signaling [32]. As shown in Fig. 6b, p-JNK (from 104–151%, p = 0.048), p-JUN (from 108–129%, p = 0.168), p-IκBα (from 107–133%, p = 0.371), p-P65 (from 103–125%, p = 0.112), p-P38 (from 113–132%, p = 0.316) and p-STAT1 (from 101–197%, p < 0.001) levels were elevated with addition of BV-2 CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP-expressing SH-SY5Y cells, although the elevation did not reach statistical significance in p-JUN, p-IκBα, p-P65 and p-P38. Treatment with LM-021 or tafamidis reduced p-JNK (101–98% vs. 151%, p = 0.03–0.021) and p-JUN (100–94% vs. 129%, p = 0.023–0.006), treatment with LMDS-1 reduced p-IκBα (92% vs. 133%, p = 0.046) and p-P65 (99% vs. 125%, p = 0.035), and treatment with LMDS-1 or tafamidis reduced p-P38 (100–92% vs. 132%, p = 0.018–0.002) and p-STAT1 (148–135% vs. 197%, p = 0.011–0.002) levels. There are no significant changes in total JNK (109–91%), JUN (105–98%), IκBα (102–96%), P65 (115–91%), P38 (109–102%), and STAT1 (193–167%) (p > 0.05) levels among BV-2 CM (+ LPS/IFN-γ)-inflamed ATXN3/Q75-GFP SH-SY5Y cells without or with compound treatment. Nevertheless, total STAT1 expression was significantly elevated upon addition of BV-2 CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP-expression cells (from 106–193%, p < 0.001).
In addition, continuous deregulated synthesis of IL-6 plays a critical role in chronic inflammation [33]. IL-6 may act via the JAK2/STAT3 pathway, and SOCS3 hinders the IL-6 signaling with high potency and specificity [34]. As shown in Fig. 6c, levels of IL-6 (from 113–173%, p = 0.017), p-JAK2 (from 92–140%, p = 0.015), and p-STAT3 (from 101–150%, p = 0.007) were enhanced with addition of BV-2 CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP-expressing SH-SY5Y cells, whereas treatment with LMDS-2 or tafamidis lowered the IL-6 (109–96% vs. 173%, p = 0.01–0.002), p-JAK2 (99–98% vs. 140%, p = 0.047–0.039) and p-STAT3 (112–105% vs. 150%, p = 0.045–0.014) levels. There are no changes in total JAK2 (109–99%) and STAT3 levels (148–143%) (p > 0.05) among BV-2 CM (+ LPS/IFN-γ)-inflamed ATXN3/Q75-GFP SH-SY5Y cells without or with compound treatment, although total STAT3 expression was significantly elevated upon addition of BV-2 CM (+ LPS/IFN-γ) to ATXN3/Q75-GFP-expression cells (from 98–148%, p = 0.001). Furthermore, the reduced SOCS3 (from 95–80%, p = 0.484) was rescued (107–114% vs. 80%, p = 0.026–0.004) by the treatment with LMDS-2 or tafamidis. Together, the results in Fig. 6 showed the anti-inflammatory effects of LM-021, LMDS-1, LMDS-2 and tafamidis on SCA3 neuronal cells.